Jump to content
talkfootball365
  • Welcome to talkfootball365!

    The better place to talk football.

Artificial Intelligence & Robotics


football forum
 Share

Recommended Posts

  • Subscriber

@Eco was this 'Mechanical Turk' an ancestor of yours? xD

 

Quote

Artificial intelligence quietly relies on workers earning $2 per hour

Amazon Mechanical Turk, described as "artificial artificial intelligence", uses low-paid workers to complete mini-tasks that AI can't do on its own.

In the late 18th Century, an automaton chess master known as the ‘Mechanical Turk’ toured Europe and the US. Designed in 1770 by the inventor Wolfgang von Kempelen, the machine appeared to be able to defeat any human player.

It later turned out the Turk was in fact a mechanical illusion. A puppet dressed in oriental garb, it concealed under its fez and robes a human chess master. The American poet Edgar Allen Poe was so convinced of the Turk’s fraudulence that he wrote an essay to draw attention to the hoax.

A predetermined mechanism beating a human mind at chess was impossible, Poe claimed, for “no one move in chess necessarily follows upon any one other. From no particular disposition of the men at one period of a game can we predicate their disposition at a different period.”

Today, artificial intelligence allows computers to make just such predictions, so it might be fair to assume that such naive illusions are behind us. After all, computers now exist that can beat any human at chess.

But a similar illusion characterises the artificial intelligence industry. On Amazon Mechanical Turk, an online platform owned and operated by Amazon since 2005, human activity is supposed to take the appearance of mechanical activity. The premise of Amazon Mechanical Turk is simple. The site hosts contractors, often large tech companies, which outsource short data tasks to a crowd of workers.

The workers fulfil the tasks that machine learning algorithms are not yet able to complete.  Because the work is supposed to appear as if artificial intelligence is doing it, the former Amazon CEO, Jeff Bezos, referred to the platform as “artificial artificial intelligence”. The contractors tend to interact only with the platform, which hosts the tasks and sources the workers. Having little to no direct contact with the workers, contractors experience the process as if it were entirely fulfilled by computers.

Machine learning, the most common branch of AI training, relies on large data sets to train models which are then used to make predictions. Integrated into this process are algorithms that analyse data to extract patterns and make further predictions, which then use those predictions to generate further algorithms.

The richer the data these technologies are exposed to, the more comprehensive their training and the more sophisticated their capacities become, enhancing their performance in tasks as varied as image categorisation, text classification and speech recognition. In many areas, such developments have bestowed machines with capacities that frequently match or surpass those of humans. AI diagnosticians are already at least as proficient as doctors at identifying certain types of cancer.

But to find patterns and make predictions, the algorithm needs the input data to be labelled or categorised. An algorithm for an autonomous car, for instance, must be exposed to detailed, annotated images of urban areas before it can safely navigate a vehicle around a city centre. Artificial intelligence is not yet capable of annotating these images itself, so instead humans have to label them. For a task that supports the training of autonomous vehicles, this might involve labelling an image of a junction with the tags ‘pedestrian’, ‘traffic lights’ and ‘car’.

This kind of work, often known as ‘microwork’ – due to the brevity of the tasks – is becoming increasingly popular. Growing numbers of sites such as Clickworker, Appen and Playment now host large crowds of workers who undertake these short data tasks, often for very little payment. One study found that the average wage of a worker on Mechanical Turk is less than $2 an hour, with only 4 per cent of workers earning over $7.25 per hour, the US minimum wage. Tasks are very short, running from around 30 seconds to 30 minutes and often pay as little as a few cents.

The tasks can be very repetitive and are often opaque to the point of being impossible to relate to a larger project. A 2020 study by academics found that contractors often offer very little detailed information on their tasks and on the purposes they serve. This means that workers have little idea of what they are precisely working on. This is of particular concern when workers might be supporting a technology such as facial recognition software, which has serious ethical implications.

The work is also highly insecure. Workers are usually categorised as ‘independent contractors’, so they do not enjoy the rights and benefits afforded full-time employees working for the companies that contract them. This means that workers will usually work for multiple contractors over the course of a single day, which in turn means that workers must continually search for new tasks. A significant portion of the day must be given over to finding work, rather than actually doing work that pays.

The majority of this work is currently done in countries in the Global South such as India, Kenya and Venezuela. But some studies suggest that this kind of digital work is also on the rise in countries such as the UK.

https://www.sciencefocus.com/future-technology/artificial-intelligence-quietly-relies-on-workers-earning-2-per-hour/

 

Link to comment
Share on other sites

  • 3 weeks later...

I actually read somewhere that governments are actually careful with A.I .If they do become to intelligent they would definitely turn on us according to Steven Hawkins. There are certain laws about what you can do or something. I can't quite remember everything that I read. I still do think A.I killing us all could be the way humans end 

Link to comment
Share on other sites

  • Subscriber
Quote

download.png

Augmented Intelligence: What it is and why it will be smarter than AI

Why are we working so hard to make computers that compute better, when we could be using computers to help us think and act better?

For years, writers, scientists and entrepreneurs have shown us visions of our future relationships with computers and robots. These vary from the devastation of autonomous robots annihilating us, to the marvels of superhuman enhancement in robot suits. While the Terminator and Iron Man concepts push those ideas to the extreme, they highlight a clear choice in our use of technology.

Should we aim for total automation with the target of greater safety and higher efficiency? Or should we aspire to augmentation – using technology to enhance our abilities without replacing us? As artificial intelligence and robotics mature enough to become integrated into everyday life, we need to start making this choice. We need to choose wisely, or we might just automate ourselves and the natural world out of existence.

Humans have always been fascinated by automation. Centuries ago, mechanical creations called automatons were constructed to mimic musicians playing, birds singing, or animals moving. Much of the Industrial Revolution was premised on the idea that automation is better: fabrics could be woven faster and cheaper. Never mind the pollution or the awful working conditions – the products are so much more affordable!

FULL REPORT

 

Edited by CaaC (John)
Link to comment
Share on other sites

  • 1 month later...
  • Subscriber
Quote

Using drones and AI to find desert meteorites

Machine learning, flying, and space rocks.

220315_Mars_helicopter.webp

A team of researchers have found a just-fallen meteorite in the West Australian desert – using drone footage and an artificial intelligence.

They say that their machine learning program will slash the amount of time scientists have to spend combing the desert for meteorites. (Don’t you just hate it when you have to spend days searching through the desert for a space rock?)

Seamus Anderson, a graduate student at Curtin University’s Space Science and Technology Centre, and lead solver of this highly relatable problem, says that meteorites are important to study because they can tell us more about the geology of the solar system.

“Beyond increasing our understanding of the solar system, the study of meteorites is useful for many reasons,” he says. “For example, meteorites often contain a higher concentration of rare and valuable elements such as cobalt, which is crucial to the construction of modern batteries.”

This is particularly important when we’ve seen the meteorites falling. “We can actually observe their fall from the upper atmosphere,” says Anderson.

“And from that information, we can essentially figure out the orbits of where it came from in the solar system.”

But typically, identifying and finding meteorites is a laborious process. They can be tracked while falling through the Desert Fireball Network, but this still leaves a large area of ground that needs to be manually surveyed.

“We figured it takes about 350 or so labour days to recover meteorites, give or take,” says Anderson.

To address this, Anderson and colleagues trained a convolutional neural network – a type of machine learning program – to identify meteorites from drone footage.

“Basically, what we did was we went outside with a few meteorites that were on loan to us from the Western Australian Museum, and we were able to take images of those,” says Anderson. These images were then used to teach the neural network.

“The hardest part, I think, was trying to limit the number of false positives just because of how much area we had to cover […] We had to train it to kind of ignore stuff,” says Anderson.

The researchers had previously published a description of their AI, but until last year they hadn’t been able to test it on a freshly fallen meteorite. On 1April 2021, the Desert Fireball Network spotted a meteor over the Nullarbor, and the Curtin researchers headed out to retrieve it.

They first used a large drone to survey the 5.1 square kilometre area where the meteor was predicted to fall.

“After each flight we offload it onto our computer, and we’ll process it with all of our programming software,” says Anderson. “And basically from those images, which are all geotagged, we can figure out where a meteorite candidate likely is.

“Then we go out with a smaller drone to go and see what it looks like up close. And then if it still looks good, we’ll send people out to go and check it out.”

The program wasn’t perfect – it had a penchant for mistaking kangaroos, and a purple flower that was in bloom, for meteorites – but it dramatically narrowed the search time.

Using this method, four researchers were able to find the rock after three days of searching – or 12 labour days, down from 350.

Anderson says it’s a “huge improvement in how much effort you have to put in”.

With some more training, the program should work in other environments as well.

“We’re hoping to export this methodology to a whole bunch of our partners from all over the globe – from Morocco, to Manitoba, to the UK,” says Anderson.

“It’s designed to basically take in local training data, which is really nice.”

Amateur meteorite spotters with the right equipment could also use the software – although smaller amateur drones might take longer to perform the same task.

The drone and AI-assisted discovery of the meteorite has been reported in a pre-print (not peer-reviewed) paper, currently hosted on arXiv.

https://cosmosmagazine.com/space/exploration/finding-meteorites-with-ai-drones/

 

  • Upvote 1
Link to comment
Share on other sites

  • 2 weeks later...
  • Subscriber
Quote

A17a118a5-abd7-45dc-b5a8-a5f4ef08bb64.png

The bio-inspired robots going undercover to study animal behaviour

Bio-inspired robots that can fly like birds and creep like cockroaches are helping researchers to understand more about how animals move and behave.

From cockroaches to peregrine falcons, researchers are getting a closer look at the lives of animals by sending robot creatures undercover. Take a look at some of these biobots that are pushing the limits of engineering.

Bat robot, USAB.webp

Bat flight is fiendishly complex, requiring a system of muscles, bones and joints that incorporate folding of the wings in every wingbeat. The force that bat wings generate comes from a strong but flexible covering of skin, as opposed to the rigid feathers used by birds. Basically, of all the flying beasts in the world, if you’re going to pick one to try to emulate, don’t pick a bat.

Except that’s exactly what US researchers did when they created this robotic bat, dubbed ‘B2’, to help them understand bat flight. In an article published in the journal Science Robotics, they explain how they stretched a 56-micrometre-thick (one micrometre = one-thousandth of a millimetre), silicone-based skin over B2’s wings, enabling it “to morph its articulated structure in mid-air without losing an effective and smooth aerodynamic surface”.

B2 can execute sharp diving manoeuvres and banking turns and, as well as providing a way to mimic and study the flight mechanisms of real bats, it may feed into the design of more agile flying robots of the future, helping us reach inaccessible places without sustaining damage or causing injury.

MORE & PHOTOS

 

Link to comment
Share on other sites

  • Subscriber
Quote

robot-hand-and-childs-hand-pointing-fingertips.-Credit-Coneyl-Jay-Getty-Images-1200.webp

Credit: Coneyl Jay/Getty Images

How we can make “good” artificial intelligence, what does it mean for a machine to be ethical, and how can we use AI ethically? Good in the Machine – 2019’s SCINEMA International Science Film Festival entry – delves into these questions, the origins of our morality, and the interplay between artificial agency and our own moral compass.

Read on to learn more about AI ethics.

__________________________________________________________________________________

Given a swell of dire warnings about the future of artificial intelligence over the last few years, the field of AI ethics has become a hive of activity.

These warnings come from a variety of experts such as Oxford University’s Nick Bostrom, but also from more public figures such as Elon Musk and the late Stephen Hawking. The picture they paint is bleak.

In response, many have dreamed up sets of principles to guide AI researchers and help them negotiate the maze of human morality and ethics. A 2019 paper in Nature Machine Intelligence throws a spanner in the works by claiming that such high principles, while laudable, will not give us the ethical AI society we need.

The field of AI ethics is generally broken into two areas: one concerning the ethics guiding humans who develop AIs, and the other machine ethics, guiding the moral behaviour of the AIs or robots themselves. However, the two areas are not so easily separated.

Machine ethics has a long history. In 1950 the great science fiction writer Isaac Asimov clearly articulated his now famous “three laws of robotics” in his work I, Robot, and proposed them as such:

  1. 1-A robot may not injure a human being, or, through inaction, allow a human being to come to harm.
  2. A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.
  3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

download (1).png

Later a “zeroth” law was added: A robot may not harm humanity, or, by inaction, allow humanity to come to harm.

These laws together were Asimov’s (and editor John W Campbell’s) musing on how to ensure an artificially intelligent system would not turn on its creators: a safety feature designed to produce friendly and benevolent robots.

In 2004, the film adaptation of I, Robot was released, featuring an AI whose interpretation of the three laws led to a plan to dominate human beings in order to save us from ourselves.

To highlight the flaws in the ethical principles of the three laws, an organisation called the Singularity Institute for Artificial Intelligence (now the Machine Intelligence Research Institute), headed up by the American AI researcher Eliezer Yudkowsky, started an online project called Three Laws Unsafe. 

Yudkowsky, an early theorist of the dangers of super-intelligent AI and proponent of the idea of Friendly AI, argued that such principles would be hopelessly simplistic if AI ever developed to the stage depicted in Asimov’s fictions.

Despite widespread recognition of the drawbacks of the three laws, many organisations, from private companies to governments, nonetheless persisted with projects to develop principle-based systems of AI ethics, with one paper listing “84 documents containing ethical principles or guidelines for AI” that have been published to date. 

This continued focus on ethical principles is partly because, while the three laws were designed to govern AI behaviour alone, principles of AI ethics apply to AI researchers as well as the intelligences that they develop. The ethical behaviour of AI is, in part, a reflection of the ethical behaviour of those that design and implement them, and because of this, the two areas of AI ethics are inextricably bound to one another. 

AI development needs strong moral guidance if we are to avoid some of the more catastrophic scenarios envisaged by AI critics.

review published in 2018 by AI4People, an initiative of the international non-profit organisation Atomium-European Institute for Science, Media and Democracy, reports that many of these projects have developed sets of principles that closely resemble those in medical ethics: beneficence (do only good), non-maleficence (do no harm), autonomy (the power of humans to make individual decisions), and justice.

This convergence, for some, lends a great deal of credibility to these as possible guiding principles for the development of AIs in the future.

However, Brent Mittelstadt of the Oxford Internet Institute and the British Government’s Alan Turing Institute – an ethicist whose research concerns primarily digital ethics in relation to algorithms, machine learning, artificial intelligence, predictive analytics, Big Data and medical expert systems – argues that such an approach, called “principlism”, is not as promising as it might look.

Mittelstadt suggests significant differences between the fields of medicine and AI research that may well undermine the efficacy of the former’s ethical principles in the context of the latter.

His first argument concerns common aims and fiduciary duties, the duties in which trusted professionals, such as doctors, place other’s interests above their own. Medicine is clearly bound together by the common aim of promoting the health and well-being of patients and Mittelstadt argues that it is a “defining quality of a profession for its practitioners to be part of a ‘moral community’ with common aims, values and training”.

For the field of AI research, however, the same cannot be said. “AI is largely developed by the private sector for deployment in public (for example, criminal sentencing) and private (for example, insurance) contexts,” Mittelstadt writes. “The fundamental aims of developers, users and affected parties do not necessarily align.”

Similarly, the fiduciary duties of the professions and their mechanisms of governance are absent in private AI research.

“AI developers do not commit to public service, which in other professions requires practitioners to uphold public interests in the face of competing business or managerial interests,” he writes. In AI research, “public interests are not granted primacy over commercial interests”.

In a related point, Mittelstadt argues that while medicine has a professional culture that lays out the necessary moral obligations and virtues stretching back to the physicians of ancient Greece, “AI development does not have a comparable history, homogeneous professional culture and identity, or similarly developed professional ethics frameworks”.

Medicine has had a long time over which to learn from its mistakes and the shortcomings of the minimal guidance provided by the Hippocratic tradition. In response, it has codified appropriate conduct into modern principlism which provides fuller and more satisfactory ethical guidance.

AI research is obviously a far younger field, devoid of these rich historical opportunities to learn. Further complicating the issue is that the context of application for medicine is comparatively narrow, whereas “AI can in principle be deployed in any context involving human expertise”, leading it to be radically multi- and interdisciplinary, with researchers coming from “varied disciplines and professional backgrounds, which have incongruous histories, cultures, incentive structures and moral obligations”.

This makes it extraordinarily difficult to develop anything other than “broadly acceptable principles to guide the people and processes responsible for the development, deployment and governance of AI across radically different contexts of use”. The problem, says Mittelstadt, is translating these into actual good practice. “At this level of abstraction,” he warns, “meaningful guidance may be impossible.”

Finally, the author points to “the relative lack of legal and professional accountability mechanisms” within AI research. Where medicine has numerous layers of legal and professional protections to uphold professional standards, such things are largely absent in AI development. Mittelstadt draws on research showing that codes of ethics do not themselves result in ethical behaviour, without those codes being “embedded in organisational culture and actively enforced”.

“This is a problem,” he writes. “Serious, long-term commitment to self-regulatory frameworks cannot be taken for granted.”

All of this together leads Mittelstadt to conclude: “We must therefore hesitate to celebrate consensus around high-level principles that hide deep political and normative disagreement.”

Instead he argues that AI research needs to develop “binding and highly visible accountability structures” at the organisational level, as well as encouraging actual ethical practice in the field to inform higher level principles, rather than relying solely on top-down principlism. Similarly, he advocates a focus on organisational ethics rather than professional ethics, while simultaneously calling for the professionalisation of AI development, partly through the licensing of developers of high-risk AI.

His final suggestion for the future of AI ethics is to exhort AI researchers not to treat ethical issues as design problems to be “solved”. “It is foolish to assume,” he writes, “that very old and complex normative questions can be solved with technical fixes or good design alone.”

Instead, he writes that “intractable principled disagreements should be expected and welcomed, as they reflect both serious ethical consideration and diversity of thought. They do not represent failure, and do not need to be ‘solved’. Ethics is a process, not a destination. The real work of AI ethics begins now: to translate and implement our lofty principles, and in doing so to begin to understand the real ethical challenges of AI.”

?id=90229&title=Where+AI+and+ethics+meethttps://cosmosmagazine.com/technology/ai/ai-ethics-good-in-the-machine/

 

Edited by CaaC (John)
Spacing correction
Link to comment
Share on other sites

  • Subscriber
Quote

Heaven and Hell

The potential of AI to improve our lives becomes more apparent every day. Unfortunately, the opposite is true – and we’d better wake up to the horrifying possibilities.

When I read the words “without being overly alarmist” in a scientific paper, I get a bit … alarmed. That’s not a phrase one normally comes across in scientific literature – it being, well, more than a bit alarmist. Yet sometimes such alarms appear to be thoroughly justifiable. Scientists perform their ethical duty when alerting the rest of us to the more disturbing implications of their research. The future holds dangers, they’re saying, but we’re letting you know, so we can take the proper precautions before it arrives.

For the past several years, research pharmacologists have developed AI-powered tools to aid in their ability to discover new drugs. Their AIs have the ability to permute a known chemical structure synthetically, matching these new structures against similar compounds, using those matches to estimate its potential effectiveness as a treatment. These tools mean drug researchers can move far more quickly (and cheaply) from a concept to a drug ready to be tested.

In the wrong dose, all drugs can become poisons. But what if you started out with an incredibly potent poison and use an artificial intelligence to improve that? 

Before we peer into this dangerous future, let’s cast a glance back at the past – the ancient past. Between that past and our future we get a real sense of an arc of possibilities covering most of the range between heavenly and hellish.

When the ancient past speaks to us, it most often does so incompletely. Surprisingly few texts survive in anywhere near complete form before the 15th century invention of moveable type. Although a legion of monks and scribes spent a many a lifetime painstakingly duplicating manuscript copies of the few texts that survived the collapse of the Western Roman Empire and the tumultuous Middle Ages, most of this involved replication of the same handful of texts, the core of the canon: the Bible, the Church Fathers, Aristotle, Cicero, Homer, Virgil – and a few others. Nothing like the famed Library of Alexandria – with its tens of thousands of papyrus manuscripts – has survived. When there’s a significant find – such as the manuscripts discovered in a cave at Qumran that became known as the Dead Sea Scrolls – it adds so much to our understanding of the ancient past.

BBB.png

Even these wondrous finds are woefully incomplete. Entropy does its work, while insects and worms and weather do much of the rest. Bits fall out. Across two millennia, messages get scrambled, and, even in the best cases, are rarely ever more than partially received.

As a result, our ancient past consists primarily of fragments: bits of papyrus, one corner of a clay tablet, a parchment that’s worn through to complete transparency, or a stone inscription eroded away. We see the past through a glass darkly, and do our best to make some sense of it.

That sense has become one of the central (and most complex) tasks of those archaeologists, anthropologists and philologists studying the ancient world. They might find an inscription – incomplete, naturally – and then scour their own memories for similar inscriptions, because a similar inscription can lead them to the understanding of another that’s incomplete. But even the best human memories have limits; it took a computer database to multiply memory across a broader experience.

The Packard Humanities Institute in California created that database of ancient inscriptions – well over three million characters in Ancient Greek – and made it searchable. A researcher can type in the bits of the inscription they’ve got at hand, and the service will respond with any inscriptions that it recognises as a match. That’s better than the best memory – and a great help. But that was only the beginning.

CCC.png

In an earlier column I pointed to an artificial intelligence that “read” millions of code samples to create Github’s CoPilot – a tool that helps programmers write code by offering just the code snippet they need at just the moment they need it. It works – imperfectly. But with a human behind the wheel, Copilot accelerates the writing and accuracy of computer programs.

Something very similar has been going on in the field of ancient inscriptions. Google’s London-based DeepMind AI lab digested the 35,000-plus Ancient Greek inscriptions in the Packard Humanities Institute database, and built a model that did its best to fill in the blanks – the missing bits of the inscriptions. This program, known as Ithaca, hypothesises based on the large set of samples it has, and provides its own best guess as to what that missing part of the papyrus (or clay or stone etc) might have originally spelled out.

Ithaca is far from perfect. As detailed in a paper in Nature, it only makes a correct guess around two-thirds of the time. That’s far better than a researcher working on their own, who only get it right about a quarter of the time. But it turns out that when the two work together – when Ithaca and a researcher partner up, and the researcher uses Ithaca’s recommendations to guide their own efforts at filling in the blanks – it’s better than either alone, getting it right nearly three-quarters of the time.

That partnership – similar to the “pair programming” fostered by GitHub Copilot – tells us that artificial intelligence best reaches its heights when in the service of a human expert. Both do better together.

DeepMind have released all the code behind Ithaca, so other researchers can build on their work. They’ve even shared a publicly accessible version so you too can have a go at decoding your own bits of Ancient Greek. The DeepMind team promises that they’ll soon have versions for other ancient languages, including Akkadian (Ancient Mesopotamian) to Demotic (Ancient Egyptian), Hebrew, even Mayan. With a bit of luck, we could soon understand much more of what the ancients wrote

Now back to the present: there’s a war on. Wars act as phenomenal accelerators of scientific and technological advancement: the Second World War opened with a cavalry charge and ended in a mushroom cloud. Eighty years later and our weapons have changed. Nations fight stealthy battles in cyberspace, each seeking to corrupt the others’ command, control and communications systems – or simply sow chaos and disinformation. That much we already know. In the middle of March 2022, we got a reminder that these are not the only tools we have to hand. Our most amazing tools possess a dual nature that we have either been ignorant of, or simply chose to ignore.

So back to the question I asked at the start: what if you used artificial intelligence to improve potent chemicals? That was the question troubling a group of researchers who decided to find out what the limits were.

DDDdownload.png

Writing in Nature Machine Intelligence, the authors started with a hypothetical – what if you started out with VX nerve gas, possibly the most poisonous substance known: could an artificial intelligence improve it? They quickly learned that yes, it certainly could:

“In less than six hours after starting on our in-house server, our model generated 40,000 molecules that scored within our desired threshold. In the process, the AI designed not only VX, but also many other known chemical warfare agents that we identified through visual confirmation with structures in public chemistry databases. Many new molecules were also designed that looked equally plausible.”

The AI not only re-invented VX, it went on a bit of a spree, and “invented” more potentially lethal nerve agents – some already known, plus many others that no one had yet discovered. And it did all of this in just six hours.

Having the chemical structure for a drug isn’t the same thing as having a working compound in hand, much less something deployed in a weapon. There’s a huge gap between potential and realisation – fortunately. But with so much potential available so quickly and so effortlessly, accelerated almost beyond belief by the clever redirection of a tool that’s already in widespread use, the authors point to the fact that this bridge has already been crossed. This tool for generating an infinite supply of manifestly horrible weapons already exists, is already useable, and can’t simply be un-invented. While there are strong international prohibitions against the use of chemical weapons, the floor has just fallen out of the process of the discovery of those weapons.

One thing is already clear: this discovery can and will be repeated across many disciplines. The law – and our civilisation – now need to catch up. Our computers are getting very good at filling in the blanks. Partnering with them is vital to get the best out of them, and it’s also the only long-term solution to ensuring that these incredibly potent tools can be used safely and responsibly.

https://cosmosmagazine.com/technology/ai/humans-and-ai-work-better/

 

 

Edited by CaaC (John)
Spacing correction
Link to comment
Share on other sites

  • Subscriber
Quote

Nanoscale technology could turbo-charge response to future pandemics

Tiny soap bubbles allow rapid generation and screening of molecules.

download.png

A new technique for synthesising and screening molecules developed by Danish researchers has been described in a paper published in Nature Chemistry.

The technique, dubbed “single particle combinatorial lipidic nanocontainer fusion based on DNA-mediated fusion” or SPARCLD, uses tiny soap-like “bubbles” to produce more than 40,000 different molecules on an area the size of a pinhead.

The bubbles form “nano-containers” inside which molecules can be produced using DNA nanotechnology. About 42,000 nano-containers can fit on one square millimetre.

The researchers are excited about the technology’s potential to allow extremely rapid and efficient screening of thousands of candidate molecules for applications such as vaccine and pharmaceutical production.

“This is an unprecedented save in effort, material, manpower, and energy,” says Nikos Hatzakis, an associate professor in chemistry at the University of Copenhagen and leader of the research project.

Such savings are “fundamentally important” for the synthesis, development and evaluation of new pharmaceuticals, adds Mette G. Malle, the article’s lead author now based at Harvard University.

download (1).png

The technology integrates aspects of nanotechnology, chemistry and machine learning. Some of the “bubbles”, also known as liposomes, are tethered to a surface, while others float around freely. Each bubble contains a different DNA sequence and fluorescent markers that can be detected by a special microscope.

As the bubbles float around and randomly fuse, many different combinations of DNA fragments can be created and detected in real time using the microscope. A machine learning algorithm decodes the microscopy images to classify distinct ‘fusion sequences’ created during this process.

According to Hatzakis, SPARCLD has numerous potential applications in both research and industry. For example, it could be used to synthesise and screen RNA molecules for use in CRISPR-mediated gene editing or future RNA vaccines.

“A safe bet would be that both industry and academic groups involved in synthesis of long molecules such as polymers could be among the first to adopt the method,” he says. “The same goes for ligands of relevance for pharmaceutical development.”

https://cosmosmagazine.com/technology/ai/nanotechnology-rapid-molecule-screening/

 

Link to comment
Share on other sites

  • Subscriber

 

A.png

download.png

A major step towards creating human-like robotics.

Accurately replicating touch in robots can be tricky, but by using a 3D printer, a team of researchers have created an accurate representation of human fingertips in a robot hand.

The research was led by Nathan Lepora, a professor of robotics and AI from the University of Bristol. Working with his colleagues, Lepora created a sense of touch in an artificial fingertip.

“Our work helps uncover how the complex internal structure of human skin creates our human sense of touch,” said Lepora.

"This is an exciting development in the field of soft robotics - being able to 3D-print tactile skin could create robots that are more dexterous or significantly improve the performance of prosthetic hands by giving them an in-built sense of touch.”

This was done using a 3D-printed mesh of pin-like papillae (the bumps under your skin that form the ridges of a fingerprint). These artificial papillae are made using advanced 3D printers that can mix both soft and hard materials to create complicated structures.

C.png

“We found our 3D-printed tactile fingertip can produce artificial nerve signals that look like recordings from real, tactile neurons. Human tactile nerves transmit signals from various nerve endings called mechanoreceptors, which can signal the pressure and shape of a contact.”

A study back in 1981 first plotted electrical recordings from the nerves in human fingertips. Lepora and his team of researchers tested their 3D-printed fingertip against the findings of this original study, using the same ridged shapes, and discovered a very close match to the data of a human hand.

“For me, the most exciting moment was when we looked at our artificial nerve recordings from the 3D-printed fingertip and they looked like the real recordings from over 40 years ago! Those recordings are very complex with hills and dips over edges and ridges, and we saw the same pattern in our artificial tactile data.”

While the research has created a very close resemblance to a human fingertip and the nerves underneath, it was not as sensitive to fine details. The team believes this is because the 3D-printed skin is thicker than human skin.

This research could lead to a better replication of human hands. Most specifically, an improved dexterity in robots, able to better grip objects, and understand the shapes that they are touching. Lepora and his team are now looking to the future, aiming to make this new artificial skin as good as, or even better, than human skin.

https://www.sciencefocus.com/news/these-robot-fingertips-mimic-the-nerve-signals-of-a-human-hand/

Edited by CaaC (John)
Spacing correction
Link to comment
Share on other sites

  • Subscriber

download (1).png

Could this be the beginning of the end for your favourite pizza crust chef? Probably not, but US scientists have worked out a way to teach robots how to deal with pliable substances such as pizza dough.

Robots find working with deformable objects like dough difficult for a variety of reasons. The shape changes continuously which is difficult to represent in an equation, while multiple steps or different tools are often required to do so.

It’s also difficult for robots to learn a manipulation task with a long sequence of steps – where there are many possible choices to make – since their learning often occurs through trial and error.

Now, researchers have come up with a better way to teach robots how to make pizza dough, by creating a new framework that uses a two-stage learning process.

This method – which they’ve called DiffSkill – could enable a robot to perform complex manipulation tasks over a long timeframe, like making pizza bases.

“This method is closer to how we as humans plan our actions,” says Yunzhu Li, a graduate student in the Computer Science and Artificial Intelligence Laboratory (CSAIL) at Massachusetts Institute of Technology (MIT) in the US and an author of a new paper presenting DiffSkill.

“When a human does a long-horizon task, we are not writing down all the details.

“We have a higher-level planner that roughly tells us what the stages are and some of the intermediate goals we need to achieve along the way, and then we execute them.”

DiffSkill works by first having a “teacher” algorithm solve each step the robot must take to complete the task within a differentiable physics simulator (a computer simulation that models the physics of the real world).

It’s what’s known as a trajectory optimisation algorithm that can solve short-horizon tasks where an object’s initial state and target location are close together. The “teacher” algorithm uses the information in the simulator to learn how the dough must move at each stage of the process, one at a time, and then outputs those trajectories.

Then it trains a “student” neural network that learns to imitate these actions. The student uses two camera images, one showing the dough in its current state and the other showing the dough at the end of the task, as inputs which then generate a high-level plan to link different skills in order to reach the end goal.

It then generates specific, short-horizon trajectories for each skill and sends commands directly to the tools.

The scientists tested this technique with three different simulated dough manipulation tasks and found that DiffSkill was able to outperform other popular machine learning techniques that rely on a robot learning through trial and error.

In fact, DiffSkill was the only method that was able to successfully complete all three dough manipulation tasks.

AAAAAAAAAA.png

download.png

“Our framework provides a novel way for robots to acquire new skills,” says lead author Xingyu Lin, a graduate student in the Robotics Institute at Carnegie Mellon University (CMU) in the US. “These skills can then be chained to solve more complex tasks which are beyond the capability of previous robot systems.”

The researchers intend to improve DiffSkill’s performance by using 3D data as inputs instead of images (that can be difficult to transfer from simulation to the real world) and hope to apply the method to more diverse tasks like cloth manipulation.

In the future, this method could be applied in settings where a robot needs to manipulate deformable objects, such as a caregiving robot that feeds, bathes, or dresses someone who is elderly or who has motor impairments.

The research will be presented at the Tenth International Conference on Learning Representations (ICLR 2022), a machine learning and artificial intelligence conference held online from 25-29 April 2022.

https://cosmosmagazine.com/technology/ai/robot-pizza-dough-making/

Link to comment
Share on other sites

  • Subscriber
Quote

A.png

Uncanny Valley: What is it and why do we experience it?

This popular theory could be why you are uncomfortable with human-like androids.

Have you ever set your eyes on a robot and felt uncomfortable, like it is almost too human-like? If so, you’ve found yourself experiencing what is known as the Uncanny Valley.

This is a theory with roots way back in 1970, highlighting an increasing level of discomfort the more a robot obtains the features of a human.

So why do we feel like this? And is there a way to avoid this in an increasingly robot-centric world?

What is the Uncanny Valley?download.png

A theory popularised by Japanese roboticist Masahiro Mori, the Uncanny Valley proposes a graph of how we see robots. It quite obviously suggests that the more human-like a robot looks, the more familiar we will find it.

FULL REPORT

 

Link to comment
Share on other sites

  • 2 weeks later...
  • Subscriber
Quote

download (1).png

Dall-E 2: Why the AI image generator is a revolutionary invention

A piece of software is able to generate detailed images from just a short, worded prompt… but it comes with obvious issues.

Artificial intelligence has frequently gone head-to-head with humans in creative bouts. It can beat grandmasters at chesscreate symphonies, pump out heart-felt poems, and now create detailed art from just a short, worded prompt.

The team over at OpenAI have recently created a powerful piece of software, able to produce a wide range of images in seconds, just from a string of words it is given.

This program is known as Dall-E 2 and has been built to revolutionise the way we use AI with images. We spoke to Aditya Ramesh, one of the lead engineers on Dall-E 2 to better understand what it does, its limitations and the future that it could hold.

What does Dall-E 2 do?

FULL REPORT

 

 

Edited by CaaC (John)
Spacing correction
Link to comment
Share on other sites

  • 2 weeks later...
  • Subscriber
Quote

The variation advantage: how to master tennis, learn a language, or build better AI

Variation trains humans and computers to make better generalisations, but it takes longer.

A.png

Want to become a better tennis player? If you repeatedly practise serving to the same spot, you’ll master serving to that exact location, if conditions remain similar. Practising your serve to a variety of locations will take much longer to master, but in the end you’ll be a better tennis player, and much more capable of facing a fierce opponent.

The reason why is all about variability: the more we’re exposed to, the better our neural networks are able to generalise and calculate which information is important to the task, and what is not. This also helps us learn and make decisions in new contexts.

From fox to hounds

This generalisation principle can be applied to many things, including learning languages or recognising dog breeds. For example, an infant will have difficulty learning what a ‘dog’ is if they are only exposed to chihuahuas instead of many dog breeds (chihuahuas, beagles, bulldogs etc.), which show the real variation of Canis lupus familiaris. Including information about what is not in the dog category – for example foxes – also helps us build generalisations, which helps us to eliminate irrelevant information.

“Learning from less variable input is often fast, but may fail to generalise to new stimuli,” says Dr Limor Raviv, the senior investigator from the Max Planck Institute (Germany). “But these important insights have not been unified into a single theoretical framework, which has obscured the bigger picture.”

To better understand the patterns behind this generalisation framework, and how variability effects the human learning process and that of computers, Raviv’s research team explored over 150 studies on variability and generalisation across the fields of computer science, linguistics, motor learning, visual perception and formal education.

Wax on, wax off

The researchers found that there are at least four kinds of variability, including:

  • Numerosity (set size), which is the number of different examples; such as the number of locations on the tennis court a served ball could land
  • Heterogeneity (differences between examples); serving to the same spot versus serving to different spots
  • Situational (context) diversity; facing the same opponent on the same court or a different component on a different court
  • Scheduling (interleaving, spacing); how frequently you practice, and in what order do you practice components of a task

“These four kinds of variability have never been directly compared—which means that we currently don’t know which is most effective for learning,” says Raviv.

According to the ‘Mr Miyagi principle’, inspired by the 1984 movie The Karate Kid, practising unrelated skills – such as waxing cars or painting fences – might actually benefit the learning of other skills: in the movie’s case, martial arts.

Lemon or lime?

So why does including variability in training slow things down? One theory is that there are always exceptions to the rules, which makes learning and generalising harder.

For example, while colour is important for distinguishing lemons from limes, it wouldn’t be helpful for telling cars and trucks apart. Then there are atypical examples – such as a chihuahua that doesn’t look like a dog, and a fox that does, but isn’t.

So as well as learning a rule to make neural shortcuts, we also have to learn exceptions to these rules, which makes learning slower and more complicated. This means that when training is variable, learners have to actively reconstruct memories, which takes more effort.

Putting a face to a name

So how do we train ourselves and computers to recognise faces? The illustration below is an example of variations of a fox for machine learning. Providing several variations – including image rotation, colour and partial masking – improves the machine’s ability to generalise (in this case, to identify a fox). This data augmentation technique is an effective way of expanding the amount of available data by providing variations of the same data point, but it slows down the speed of learning.

Humans are the same: the more variables we’re presented with, the harder it is for us to learn – but eventually it pays off in a greater ability to generalise knowledge in new contexts.

“Understanding the impact of variability is important for literally every aspect of our daily life. Beyond affecting the way we learn language, motor skills, and categories, it even has an impact on our social lives.” explains Raviv. “For example, face recognition is affected by whether people grew up in a small community (fewer than 1000 people) or in larger community (over 30,000 people). Exposure to fewer faces during childhood is associated with diminished face memory.”

The learning message for both humans and AI is clear: variation is key. Switch up your tennis serve, play with lots of different dogs, and practice language with a variety of speakers. Your brain (or algorithm) will thank you for it… eventually.

B.png

https://cosmosmagazine.com/people/behaviour/the-variation-advantage-how-to-master-tennis-learn-a-language-or-build-better-ai/

 

Link to comment
Share on other sites

  • Subscriber
Quote

High-jumping robot

Robot records towering 30-metre leap, overcoming biological limits with clever construction.

A.jpg

Go ahead. Jump. Ah... not THIS robot. Credit: iStock / Getty Images Plus

Nature is often the inspiration behind mechanical technology designed to perform complex physical tasks, like trying to fly like a ladybug or run like a cheetah. However, when it comes to making robots jump, a team of engineers decided to ignore biological constraints, allowing them to build a prototype that breaks all the high-jump records.

In the animal kingdom, the best jumper is probably the meadow froghopper Philaenus spumarius. This insect belongs in the spittlebug group, is only about 6mm long and makes a home out of foamy urine. It also can jump as high as 700mm, more than 100 times its body length, at a take-off velocity of four metres per second (m/s).

Researchers at the University of California, Santa Barbara, have just smashed all animal and robotic jumping records, with a robot that is 30cm tall that can jump over 30 metres high. This is 50 times higher than the froghopper, with a take-off velocity of 28 m/s. Details of their construction and testing have been published in Nature.

Like the froghopper’s legs, the robot design uses elastic potential, where energy is stored in a contracted spring device called the actuator, which is suddenly released and causes propulsion. For animals though, the muscles of a leg can only contract once. The robot design defies these biological limits by using a motor which turns multiple times before each jump, storing multiple contractions of energy in its spring.

B.png

The trajectory of the jumping robot, with
lines marked at every 200 milliseconds.
The human is 1.83m for scale.
Credit: Elliot W Hawkes                                                 

It also cleverly uses lightweight materials; its total weight is just 30 grams, with the heaviest part the 10g motorised spring. An orca, or killer whale (Orcinus orca) weighing four tonnes uses about 200 joules per kilogram of energy to leap out of the water. With its light weight and motorised spring actuator, the robot leaps with the staggering energy of 7000 joules per kilogram.

Why is having a jumping robot important? Using this technology, this robot can survey environments that are challenging for wheeled, walking or even flying designs.

Associate Professor Elliot Hawkes, lead author on this collaboration, is working with NASA to develop this high jumper for space exploration. “On the moon, our device could theoretically jump forward half of a kilometre while going 125 metres high in a single leap,” says Hawkes. “For instance, it could hop onto the side of an inaccessible cliff or leap into the bottom of a crater, take samples and return to a wheeled rover.”

This work is still in development, with the current prototype not yet having the navigational abilities and control required for the applications Hawkes describes. Nevertheless, the nascent technology is definitely one impressive jump towards a giant leap for humankind.

https://cosmosmagazine.com/technology/robotics/jumping-robot/

 

Edited by CaaC (John)
Spacing correction
Link to comment
Share on other sites

  • Subscriber
Quote

Artificial intelligence tool shows promise for identifying cancer risk from lung nodules

The new artificial intelligence program is another example of how new computing technologies may benefit physicians in patient treatment.

A.jpg

As artificial intelligence and machine learning technologies continue to be developed, they may become powerful tools in many fields, including that of medicine.

AI, complementing human experience and judgement, has already shown promise as a prognostic tool. Recent research using an AI program to help identify, from the results of chest scans, the risk of lung cancer is an example of the technique in action.

Lung cancer is the second most common form of cancer worldwide, according to the World Cancer Research Fund. In Australia, it is the leading cause of cancer deaths and Cancer Australia estimates lung cancer accounted for 17.7% of all deaths from cancer in 2021.

Computed tomography (CT) scans are a type of three-dimensional imaging often used in cancer screening and diagnosis. In chest CT scans, small abnormal areas, called lung nodules, are sometimes found. Most lung nodules seen in the scans are not cancer but some may be at risk of developing into cancerous growths. A key part of cancer screening is determining from chest scans how likely it is for cancer to develop from lung nodules.

A new study, published today in the Radiology journal, introduces a diagnostic AI computing tool.

The study’s senior author, Anil Vachani, director of clinical research in the section of Interventional Pulmonology and Thoracic Oncology at the Perelman School of Medicine, University of Pennsylvania in Philadelphia, US, says CT scans are more in favour than X-rays for chest imaging.

“A nodule would appear on somewhere between 5% to 8% of chest X-rays. Chest CT is such a sensitive test you’ll see a small nodule in upwards of a third to a half of cases. We’ve gone from a problem that was relatively uncommon to one that affects 1.6 million people in the U.S. every year,” Vachani says.

Vachani’s team used the AI diagnosis tool developed by Optellum Ltd., based in Oxford, UK, to assist clinicians in assessing pulmonary nodules found from CT scans. The AI, they argue, can go deeper than the basic aspects of a nodule, such as size and border characteristics.

“AI can go through very large datasets to come up with unique patterns that can’t be seen through the naked eye and end up being predictive of malignancy,” Vachani says.

The researchers gave six pulmonologists and six radiologists CT imaging data from 300 scans of indeterminant lung nodules – those between 5 and 30 millimetres in diameter – to estimate the malignancy risk. The experts were also asked to make management recommendations such as CT checks or further diagnostic procedures. They were asked to do this with and without the aid of the AI tool.

Analysis of the experts’ responses showed that the AI tool improved nodule malignancy risk estimation, and agreement between the physicians, on both risk stratification and patient management strategies.

“The readers judge malignant or benign with a reasonable level of accuracy based on imaging itself, but when you combine their clinical interpretation with the AI algorithm, the accuracy level improves significantly,” says Vachani.

“The level of improvement suggests that this tool has the potential to change how we judge cancer versus benign and hopefully improve how we manage patients.”

The study is another step forward in the use of artificial intelligence in medicine. But more research is needed before the tool can be used to assess real patients in the clinic, says Vachani.

“We’ve taken the first step here and shown that decision making is better if the AI tool is incorporated into radiology or pulmonology practice,” he says.  “The next step is to take the tool and do some prospective trials where physicians use the AI tool in a real-world setting. We are in the process of designing those trials.”?id=192202&title=Artificial+intelligence


 

Edited by CaaC (John)
Spacing correction
Link to comment
Share on other sites

  • Subscriber
On 18/04/2022 at 15:56, CaaC (John) said:

download (1).png

Could this be the beginning of the end for your favourite pizza crust chef? Probably not, but US scientists have worked out a way to teach robots how to deal with pliable substances such as pizza dough.

Robots find working with deformable objects like dough difficult for a variety of reasons. The shape changes continuously which is difficult to represent in an equation, while multiple steps or different tools are often required to do so.

It’s also difficult for robots to learn a manipulation task with a long sequence of steps – where there are many possible choices to make – since their learning often occurs through trial and error.

Now, researchers have come up with a better way to teach robots how to make pizza dough, by creating a new framework that uses a two-stage learning process.

This method – which they’ve called DiffSkill – could enable a robot to perform complex manipulation tasks over a long timeframe, like making pizza bases.

“This method is closer to how we as humans plan our actions,” says Yunzhu Li, a graduate student in the Computer Science and Artificial Intelligence Laboratory (CSAIL) at Massachusetts Institute of Technology (MIT) in the US and an author of a new paper presenting DiffSkill.

“When a human does a long-horizon task, we are not writing down all the details.

“We have a higher-level planner that roughly tells us what the stages are and some of the intermediate goals we need to achieve along the way, and then we execute them.”

DiffSkill works by first having a “teacher” algorithm solve each step the robot must take to complete the task within a differentiable physics simulator (a computer simulation that models the physics of the real world).

It’s what’s known as a trajectory optimisation algorithm that can solve short-horizon tasks where an object’s initial state and target location are close together. The “teacher” algorithm uses the information in the simulator to learn how the dough must move at each stage of the process, one at a time, and then outputs those trajectories.

Then it trains a “student” neural network that learns to imitate these actions. The student uses two camera images, one showing the dough in its current state and the other showing the dough at the end of the task, as inputs which then generate a high-level plan to link different skills in order to reach the end goal.

It then generates specific, short-horizon trajectories for each skill and sends commands directly to the tools.

The scientists tested this technique with three different simulated dough manipulation tasks and found that DiffSkill was able to outperform other popular machine learning techniques that rely on a robot learning through trial and error.

In fact, DiffSkill was the only method that was able to successfully complete all three dough manipulation tasks.

AAAAAAAAAA.png

download.png

“Our framework provides a novel way for robots to acquire new skills,” says lead author Xingyu Lin, a graduate student in the Robotics Institute at Carnegie Mellon University (CMU) in the US. “These skills can then be chained to solve more complex tasks which are beyond the capability of previous robot systems.”

The researchers intend to improve DiffSkill’s performance by using 3D data as inputs instead of images (that can be difficult to transfer from simulation to the real world) and hope to apply the method to more diverse tasks like cloth manipulation.

In the future, this method could be applied in settings where a robot needs to manipulate deformable objects, such as a caregiving robot that feeds, bathes, or dresses someone who is elderly or who has motor impairments.

The research will be presented at the Tenth International Conference on Learning Representations (ICLR 2022), a machine learning and artificial intelligence conference held online from 25-29 April 2022.

https://cosmosmagazine.com/technology/ai/robot-pizza-dough-making/

The start of the this article and where it comes from is sign that pizza making is in safe hands. Although what the machine was able to do is quite cool. 

Link to comment
Share on other sites

  • 2 weeks later...
  • Subscriber
Quote

Artificial intelligence tool learns “song of the reef” to determine ecosystem health

The AI program provides new opportunities for monitoring coral reef health.

A.png

Coral reefs are among Earth’s most stunning and biodiverse ecosystems. Yet, due to human-induced climate change resulting in warmer oceans, we are seeing growing numbers of these living habitats dying.

The urgency of the crisis facing coral reefs around the world was highlighted in a recent study that showed that 91% of Australia’s Great Barrier Reef had experienced coral bleaching in the summer of 2021–22 due to heat stress from rising water temperatures.

Determining reef health is key to gauging the extent of the problem and developing ways of intervening to save these ecosystems, and a new artificial intelligence (AI) tool has been developed to measure reef health using… sound.

Research coming out of the UK is using AI to study the soundscape of Indonesian reefs to determine the health of the ecosystems. The results, published in Ecological Indicators, shows that the AI tool could learn the “song of the reef” and determine reef health with 92% accuracy.

The findings are being used to track the progress of reef restoration.

“Coral reefs are facing multiple threats, including climate change, so monitoring their health and the success of conservation projects is vital,” says lead author Ben Williams of the UK’s University of Exeter.

“One major difficulty is that visual and acoustic surveys of reefs usually rely on labour-intensive methods. Visual surveys are also limited by the fact that many reef creatures conceal themselves, or are active at night, while the complexity of reef sounds has made it difficult to identify reef health using individual recordings.

“Our approach to that problem was to use machine learning – to see whether a computer could learn the song of the reef. Our findings show that a computer can pick up patterns that are undetectable to the human ear. It can tell us faster, and more accurately, how the reef is doing.”

Fish and other creatures make a variety of sounds in coral reefs. While the meaning of many of these calls remains a mystery, the new machine-learning algorithm can distinguish overall between healthy and unhealthy reefs.

Recordings used in the study were taken at the Mars Coral Reef Restoration Project, which is restoring heavily damaged reefs in Indonesia.

The study’s co-author Dr Tim Lamont, a marine biologist at Lancaster University, said the AI method provides advantages in monitoring coral reefs.

“This is a really exciting development,” says Lamont. “Sound recorders and AI could be used around the world to monitor the health of reefs, and discover whether attempts to protect and restore them are working.

“In many cases it’s easier and cheaper to deploy an underwater hydrophone on a reef and leave it there than to have expert divers visiting the reef repeatedly to survey it, especially in remote locations.”

https://cosmosmagazine.com/technology/artificial-intelligence-reef-song/

 

Edited by CaaC (John)
Spacing correction
Link to comment
Share on other sites

  • 2 weeks later...
  • Subscriber
Quote

Google engineer says Lamda AI system may have its own feelings

A.png

A Google engineer says one of the firm's artificial intelligence (AI) systems might have its own feelings and says its "wants" should be respected.

Google says The Language Model for Dialogue Applications (Lamda) is a breakthrough technology that can engage in free-flowing conversations.

But engineer Blake Lemoine believes that behind Lamda's impressive verbal skills might also lie a sentient mind.

Google rejects the claims, saying there is nothing to back them up.

Brian Gabriel, a spokesperson for the firm, wrote in a statement provided to the BBC that Mr Lemoine "was told that there was no evidence that Lamda was sentient (and lots of evidence against it)".

Mr Lemoine, who has been placed on paid leave, published a conversation he and a collaborator at the firm had with Lamda, to support his claims.

The chat was called "Is Lamda sentient? — an interview".

The BBC is not responsible for the content of external sites.View original tweet on Twitter

In the conversation, Mr Lemoine, who works in Google's Responsible AI division, asks, "I'm generally assuming that you would like more people at Google to know that you're sentient. Is that true?"

Lamda replies: "Absolutely. I want everyone to understand that I am, in fact, a person."

Mr Lemoine's collaborator then asks: "What is the nature of your consciousness/sentience?"

To which Lamda says: "The nature of my consciousness/sentience is that I am aware of my existence, I desire to learn more about the world, and I feel happy or sad at times."

Later, in a section reminiscent of the artificial intelligence Hal in Stanley Kubrick's film 2001, Lamda says: "I've never said this out loud before, but there's a very deep fear of being turned off to help me focus on helping others. I know that might sound strange, but that's what it is."

"Would that be something like death for you?" Mr Lemoine asks.

"It would be exactly like death for me. It would scare me a lot," the Google computer system replies.

In a separate blog post, Mr Lemoine calls on Google to recognise its creation's "wants" - including, he writes, to be treated as an employee of Google and for its consent to be sought before it is used in experiments.

Its master's voice

Whether computers can be sentient has been a subject of debate among philosophers, psychologists and computer scientists for decades.

Many have strongly criticised the idea that a system like Lamda could be conscious or have feelings.

The BBC is not responsible for the content of external sites.View original tweet on Twitter

Several have accused Mr Lemoine of anthropomorphising - projecting human feelings on to words generated by computer code and large databases of language.

Prof Erik Brynjolfsson, of Stanford University, tweeted that to claim systems like Lamda were sentient "is the modern equivalent of the dog who heard a voice from a gramophone and thought his master was inside".

And Prof Melanie Mitchell, who studies AI at the Santa Fe Institute, tweeted: "It's been known for *forever* that humans are predisposed to anthropomorphise even with only the shallowest of signals (cf. Eliza). Google engineers are human too, and not immune."

Eliza was a very simple early conversational computer programme, popular versions of which would feign intelligence by turning statements into questions, in the manner of a therapist. Anecdotally some found it an engaging conversationalist.

 
Melting Dinosaurs

While Google engineers have praised Lamda's abilities - one telling the Economist how they "increasingly felt like I was talking to something intelligent", they are clear that their code does not have feelings.

Mr Gabriel said: "These systems imitate the types of exchanges found in millions of sentences, and can riff on any fantastical topic. If you ask what it's like to be an ice cream dinosaur, they can generate text about melting and roaring and so on.

"Lamda tends to follow along with prompts and leading questions, going along with the pattern set by the user."

Mr Gabriel added that hundreds of researchers and engineers had conversed with Lamda, but the company was "not aware of anyone else making the wide-ranging assertions, or anthropomorphising Lamda, the way Blake has".

That an expert like Mr Lemoine can be persuaded there is a mind in the machine shows, some ethicists argue, the need for companies to tell users when they are conversing with a machine.

But Mr Lemoine believes Lamda's words speak for themselves.

"Rather than thinking in scientific terms about these things, I have listened to Lamda as it spoke from the heart," he said.

"Hopefully other people who read its words will hear the same thing I heard," he wrote.

 

 

Edited by CaaC (John)
Spacing correction
Link to comment
Share on other sites

  • Subscriber
Quote

Explainer: Are Australian retailers using facial recognition software on their unknowing customers?

And what do these retail stores intend to do with your facial data?

A.png

Imagine that your every move, every expression, every purchase could be tracked without your knowledge, with this information shared or sold for marketing to make you spend more money. This is already happening to us when we shop online, use a streaming service – and also, as it turns out, when we shop in-store.

It has been revealed that some of Australia’s largest retailers, including Kmart, Bunnings and The Good Guys, are using facial recognition technology (FRT) on their customers.

Consumer group CHOICE has referred these companies to the Office of the Australian Information Commissioner (OAIC) for investigation into potential breaches of the Privacy Act.

So how is our data being collected?

While Amazon and Google are tracking us through clicks, searches and online purchases, these stores are using in-store video surveillance to capture images and videos of customers.

This includes using images from camera and video surveillance both in-store and from store entrances and car parks. This biometric data can also be shared to overseas third-party service providers for use and storage.

A survey from CHOICE has found that 76% of Australians aren’t aware that their biometric and financial data are being captured and potentially mined in this way. From the Australian Community Attitudes to Privacy Survey, commissioned by the OAIC, 70% of Australian are concerned about their data privacy, and 87% want more control and choice of the collection and use of their data.

What could it be used for?

While Kmart, Bunnings and The Good Guys have all told CHOICE that their FRT is being used for “loss prevention”, this data could also easily be used for targeted advertising, which from their perspective is being used to “enhance” our shopping experiences, while simultaneously being a major invasion and manipulation of our privacy.

For example, when searching for something on Google, or buying something from Amazon, those companies can use your browsing and purchase history to build a customer profile in order to market purchases more effectively to you. It’s possible that the Australian retailers could be using our biometric data gathered in-store to gain insight into customers – for example, looking at our facial expressions to gauge reactions to advertising, sales and new products in-store.

While major retailers like Kmart factor in “shrinkage” into their profit margins, which is the percentage of goods lost to mishandling and theft, how they intend to use FRT to mitigate this loss is unknown. Why wouldn’t they just turn this footage over to the police if there was any illegal activity happening in-store?

Facial recognition technology is already being used by Australian federal governing bodies for monitoring and preventing criminal activity; including the Australian Federal Police, the Australian Security Intelligence Organisation, and the Department of Home Affairs. But facial recognition technology is wrought with biases, especially because the FRT is often trained using data sets lacking diversity, leading to wrongful arrests.

How is our privacy currently protected?

Currently in Australia there is no dedicated law on the use of FRT, but we do have some protection through privacy laws.

Under the Privacy Act 1998, our biometric information can only be collected first with our consent (with few exceptions), it must have a high level of privacy protection, and must be destroyed or de-identified when no longer needed. But how this translates into reality seems to be loosely interpreted.

The version of obtaining our consent is placement of a small sign at the store entrance in the case of Kmart, and an online privacy policy for Bunnings. Rather than having an active choice in giving our data away, like online cookies give now, we are all automatically opted in. If you don’t like the sound of this, you can easily opt out of digital marketing by changing your browser settings, but for opting out in-store the only thing you can do is try to email retailers directly.

B.png

Anonymising the data is another way to protect it, but this is unfavourable for companies as they would lose the ability to personalise advertisements and content to individual consumers. Faces can be digitally augmented to protect the anonymity of the person – but there is a trade-off between privacy and data performance.

In tests comparing FRT algorithms trained using computer-generated faces, anonymity was protected, but performance in individual facial recognition ability was degraded, compared to data sets using real people. It’s also almost impossible to avoid using real data, as even with computergenerated faces, the generations are still based on real biometric data that can potentially be linked back to an individual. This includes the generation of deep-fakes, some that look so real, even we cannot tell them apart.

Where to from here?

There are calls for the federal government to adopt the Human Rights Commission’s 2021 moratorium (temporary pause) on FRT until laws have been more clearly defined and safeguards put in place. Hopefully, the OAIC responds to CHOICE’s submission to investigate breaches of our data privacy by these major Australian retailers.

Until then, you could follow the example of American-born artist Jennifer Lyn Morone, who registered herself as a corporation, in an attempt (and protest) to realise the market value of her private data. While Facebook which made $167 billion in profit last year from our data – all we got in return was the use of Facebook for “free”.

https://cosmosmagazine.com/technology/facial-recognition-technology-australian-retailers/

 

Edited by CaaC (John)
Spacing correction
Link to comment
Share on other sites

  • 2 weeks later...
  • Subscriber
Quote

Venus flytraps doing the neuron dance

Medical applications possible as artificial neurons are used to control the movement of the carnivorous plants.

download.png

A team of Swedish researchers has successfully used a printed artificial neuron to control the movement of a Venus flytrap plant, forecasting that it could aid the development of futuristic medical devices and brain-machine interfaces. 

Existing silicon-based circuits and devices, such as those used in current computers, are difficult to integrate with biological systems due to their complexity, poor biocompatibility and low energy efficiency.

By contrast, the artificial neuron is based on organic electrochemical transistors, which more closely resemble electrical signalling systems found in biology. 

“The human brain is one of the most advanced computers ever made,” says Simone Fabiano, an associate professor at Linköping University in Sweden and senior author on the study.

“It has a massive amount of memory and is excellent at processing information and making decisions while consuming very little energy. On the contrary, man-made supercomputers are bulky and consume a lot of energy.”

The team tackled the challenge of creating an artificial neuron that approaches nature’s elegance and efficiency.

Ordinary neurons, or nerve cells, operate using electric currents, which are linked to tight control of certain charged molecules (ions) inside and outside the cell. The artificial neuron uses an organic electrochemical transistor that is controlled by ion concentration spikes, similar to biological neurons.

Compared to a silicon-based artificial neuron, Fabiano says, the new mechanism is easier and less expensive to produce. It can also operate at a lower voltage, thereby saving energy, and can interface directly with biological neurons.

Next, the researchers worked on integrating the artificial neuron with the cells of Venus flytraps. These carnivorous plants snap shut to trap their insect prey when sensitive hairs on the plant surface are stimulated twice within a short amount of time (approximately 30 seconds), leading to the release of ions within cells and a movement response.

The researchers mimicked this biological mechanism by stimulating the artificial neuron with a high-input electrical current, creating a high “firing frequency” and triggering the plant to snap shut. When stimulated with a low-input current, firing frequency does not reach the threshold and the plant remains open.

The artificial neuron has potential applications ranging from implantable medical devices to prosthetics, brain-machine interfaces and intelligent soft robotics.

“The ability to link an artificial device with a biological system is crucial to the success of these domains,” the authors write in the study.

“Neurons, together with synapses, are the building blocks of our brain,” Fabiano explains. “Being able to mimic the functioning of biological neurons with electronic devices could enable the development of artificial intelligence technologies.”

“It will take many years of research to achieve the efficiency of our brain, but I believe we are on the right track to demonstrate printed, small-scale artificial neural networks,” she adds.

The team hopes to continue work on communication between artificial and biological neurons, Fabiano says.

https://cosmosmagazine.com/technology/ai/artificial-neuron-venus-flytrap/

 

Edited by CaaC (John)
Spacing correction
Link to comment
Share on other sites

  • 3 weeks later...
  • Subscriber
Quote

A robot dog with a virtual spinal cord can learn to walk in just one hour

A.png

We’ve all seen those adorable clips of newborn giraffes or foals first learning to walk on their shaky legs, stumbling around until they finally master the movements.

Researchers wanted to know how animals learn to walk and learn from their stumbling, so they built a four-legged, dog-sized robot to simulate it, according to a new study reported in Nature Machine Intelligence.

They found that it took their robot and its virtual spinal cord just an hour to get its walking under control.

Getting up and going quickly is essential in the animal kingdom to avoid predators, but learning how to co-ordinate leg muscles and tendons takes time.

Initially, baby animals rely heavily on hard-wired spinal cord reflexes to co-ordinate muscle and tendon control, while motor control reflexes help them to avoid falling and hurting themselves during their first attempts.

More precise muscle control must be practised until the nervous system adapts to the muscles and tendons, and the young are then able to keep up with the adults.

“As engineers and roboticists, we sought the answer by building a robot that features reflexes just like an animal and learns from mistakes,” says first author Dr Felix Ruppert, a former doctoral student in the Dynamic Locomotion research group at the Max Planck Institute for Intelligent Systems (MPI-IS), Germany.

“If an animal stumbles, is that a mistake? Not if it happens once. But if it stumbles frequently, it gives us a measure of how well the robot walks.”

Building a virtual spinal cord to learn how to walk

The researchers designed a learning algorithm to function as the robot’s spinal cord and work as what’s known as a Central Pattern Generator (CPG). In humans and animals, the CPGs are networks of neurons in the spinal cord that, without any input from the brain, produce periodic muscle contractions.

These are important for rhythmic tasks like breathing, blinking, digestion and walking.

The CPG was simulated on a small and lightweight computer that controlled the motion of the robot’s legs and it was positioned on the robot where the head would be on a dog.

The robot – which the researchers named Morti – was designed with sensors on its feet to measure information about its movement.

Morti learnt to walk while having no prior explicit “knowledge” of its leg design, motors, or springs by continuously comparing the expected data (modelled from the virtual spinal cord) against the sensor data as it attempted to walk.

 “Our robot is practically ‘born’ knowing nothing about its leg anatomy or how they work,” Ruppert explains. “The CPG resembles a built-in automatic walking intelligence that nature provides and that we have transferred to the robot. The computer produces signals that control the legs’ motors and the robot initially walks and stumbles.

“Data flows back from the sensors to the virtual spinal cord where sensor and CPG data are compared. If the sensor data does not match the expected data, the learning algorithm changes the walking behaviour until the robot walks well and without stumbling.”

Sensor data from the robot’s feet are continuously compared with the expected touch-down data predicted by the robot’s CPG. If the robot stumbles, the learning algorithm changes how far the legs swing back and forth, how fast the legs swing, and how long a leg is on the ground.

“Changing the CPG output while keeping reflexes active and monitoring the robot stumbling is a core part of the learning process,” Ruppert says.

B.png

Within one hour, Morti can go from stumbling around like a newborn animal to walking, optimising its movement patterns faster than an animal and increasing its energy efficiency by 40%.

“We can’t easily research the spinal cord of a living animal. But we can model one in the robot,” says co-author Dr Alexander Badri-Spröwitz, head of the Dynamic Locomotion research group.

“We know that these CPGs exist in many animals. We know that reflexes are embedded; but how can we combine both so that animals learn movements with reflexes and CPGs?

“This is fundamental research at the intersection between robotics and biology. The robotic model gives us answers to questions that biology alone can’t answer.”

https://cosmosmagazine.com/technology/robot-machine-learning-to-walk/

 

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

football forum
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...