Chinese researchers published 850 papers pertaining to artificial general intelligence (AGI) between 2018 and 2022, indicating Beijing's efforts to create a thinking machine are real and active possibly including research on brain/computer interfaces.

So says think tank the Center for Security and Emerging Technology (CSET) in a recently released report that claims Beijing's effort "challenges emerging global norms, underscoring the need for a serious open source monitoring program to serve as a foundation for outreach and mitigation."

The report is based on examination of scientific papers touching on a dozen relevant technologies an effort that found 500 items of research concerning "routine AI applications" plus "a significant body of research on AGI precursor technologies, indicating that China's claims to be working toward artificial general intelligence are genuine and must be taken seriously."

Those claims were first articulated in 2017, when China published a document titled the "New Generation Artificial Intelligence Development Plan". One of the goals of that plan is "to build China's first-mover advantage in the development of AI."

The CSET report suggests that plan is advancing nicely, identifies the universities that have made the biggest contributions to the AGI effort, and notes that five of the most prolific sources of AGI research are institutions located in the city of Beijing.

But that may not mean Beijing-based boffins are doing all the heavy lifting.

"While acknowledging the Beijing-area concentration, given AGI's multi-disciplinary basis and the multiple paths through which it may be realized, the possibility of breakthroughs elsewhere in China cannot be ruled out," the report states. "Limited data suggest that Beijing may be serving as China's AGI research hub for testing and deployment done elsewhere in China, in Wuhan especially."

The report asserts "China appears to be exploring multiple paths to AGI, including a potential approach not covered in this study, namely, cognitive sharing through BCIs."

BCIs are brain/computer interfaces a meeting of wetware and hardware.

The report notes that some of China's AGI research involves researchers from overseas, but "data show that the bulk of it is situated in Chinese institutions." The authors appear to worry that it's therefore hard for the rest of the world to understand China's achievements or lack thereof and note that "hiding scientific research, for example, by restricting access to academic journals, may lead to false assumptions that devolve into a vicious cycle of measures and countermeasures."

To demonstrate the problems associated with that situation, the authors refer to the "missile gap" the mistaken belief in the 1960s that the Soviet Union's missile tech considerably exceeded that of the United States, leading to poorly-informed policy responses that did little to ease the tensions of the time.

"Pursuing this train of thought further, it is highly likely that one's inability to gauge the status and intent of a potential rival through open sources will lead to clandestine efforts to procure this same information and more driving science further underground to no-one's long-term benefit," the authors assert.

The report concludes with a call for US policymakers contemplating AGI safeguards to recognize that rivals understand the strategic significance of the tech, and of developing it first.

"Although an unrestrained race to the top is risky, unilateral restrictions on AGI development, trust-based agreements that cannot be verified, and one-sided adherence to ELSI/ELSA (ethics, legal and social implications/aspects) protocols are risky as well."

Go here to see the original:

China striving to be first source of artificial general intelligence, says think tank - The Register

Artificial intelligence (AI)

Exclusive: Prof Stuart Russell says technology could result in fewer teachers being employed possibly even none

Recent advances in AI are likely to spell the end of the traditional school classroom, one of the worlds leading experts on AI has predicted.

Prof Stuart Russell, a British computer scientist based at the University of California, Berkeley, said that personalised ChatGPT-style tutors have the potential to hugely enrich education and widen global access by delivering personalised tuition to every household with a smartphone. The technology could feasibly deliver most material through to the end of high school, he said.

Education is the biggest benefit that we can look for in the next few years, Russell said before a talk on Friday at the UNs AI for Good Global Summit in Geneva. It ought to be possible within a few years, maybe by the end of this decade, to be delivering a pretty high quality of education to every child in the world. Thats potentially transformative.

However, he cautioned that deploying the powerful technology in the education sector also carries risks, including the potential for indoctrination.

Russell cited evidence from studies using human tutors that one-to-one teaching can be two to three more times effective than traditional classroom lessons, allowing children to get tailored support and be led by curiosity.

Oxford and Cambridge dont really use a traditional classroom they use tutors presumably because its more effective, he said. Its literally infeasible to do that for every child in the world. There arent enough adults to go around.

OpenAI is already exploring educational applications, announcing a partnership in March with an education nonprofit, the Khan Academy, to pilot a virtual tutor powered by ChatGPT-4.

This prospect may prompt reasonable fears among teachers and teaching unions of fewer teachers being employed possibly even none, Russell said. Human involvement would still be essential, he predicted, but could be drastically different from the traditional role of a teacher, potentially incorporating playground monitor responsibilities, facilitating more complex collective activities and delivering civic and moral education.

We havent done the experiments so we dont know whether an AI system is going to be enough for a child. Theres motivation, theres learning to collaborate, its not just Can I do the sums? Russell said. It will be essential to ensure that the social aspects of childhood are preserved and improved.

The technology will also need to be carefully risk-assessed.

Hopefully the system, if properly designed, wont tell a child how to make a bioweapon. I think thats manageable, Russell said. A more pressing worry is the potential for hijacking of software by authoritarian regimes or other players, he suggested. Im sure the Chinese government hopes [the technology] is more effective at inculcating loyalty to the state, he said. I suppose wed expect this technology to be more effective than a book or a teacher.

Russell has spent years highlighting the broader existential risks posed by AI, and was a signatory of an open letter in March, signed by Elon Musk and others, calling for a pause in an out-of-control race to develop powerful digital minds. The issue has become more urgent since the emergence of large language models, Russell said. I think of [artificial general intelligence] as a giant magnet in the future, he said. The closer we get to it the stronger the force is. It definitely feels closer than it used to.

Policymakers are belatedly engaging with the issue, he said. I think the governments have woken up now theyre running around figuring out what to do, he said. Thats good at least people are paying attention.

However, controlling AI systems poses both regulatory and technical challenges, because even the experts dont know how to quantify the risks of losing control of a system. OpenAI announced on Thursday that it would devote 20% of its compute power to seeking a solution for steering or controlling a potentially super-intelligent AI, and preventing it from going rogue.

The large language models in particular, we have really no idea how they work, Russell said. We dont know whether they are capable of reasoning or planning. They may have internal goals that they are pursuing we dont know what they are.

Even beyond direct risks, systems can have other unpredictable consequences for everything from action on climate change to relations with China.

Hundreds of millions of people, fairly soon billions, will be in conversation with these things all the time, said Russell. We dont know what direction they could change global opinion and political tendencies.

We could walk into a massive environmental crisis or nuclear war and not even realise why its happened, he added. Those are just consequences of the fact that whatever direction it moves public opinion, it does so in a correlated way across the entire world.

{{topLeft}}

{{bottomLeft}}

{{topRight}}

{{bottomRight}}

{{.}}

Read more here:

AI likely to spell end of traditional school classroom, leading expert says - The Guardian

Tom Cruise is back in Mission: Impossible Dead Reckoning, busting out lots of nifty-sounding technologies. But do any of them actually exist? (Paramount Pictures via AP)

Mission: Impossible films are always a lot of fun, and the seventh, latest instalment Mission: Impossible Dead Reckoning Part 1, which came out in UK cinemas yesterday, is just as engaging and exhilarating as the previous films, not to mention incredibly timely, given recent news events.

Tom Cruise returns as Ethan Hunt, who in 2023 is still completing missions for the Impossible Mission Force (IMF) no, not the International Monetary Fund, the other one but this time he faces off against a mysterious, sentient, all-powerful artificial intelligence (AI) intent on controlling the world, which calls itself the Entity.

But how does the technology in MI7 stack up? Heres the down-low from The Standard and some technical experts.

Warning: Contains spoilers

Sentient AI, also known as The Entity, as depicted in Mission: Impossible Dead Reckoning Part 1 (Paramount Pictures)

In MI7, someone has built a terrifying new weapon that threatens all humanity a godless, stateless, amoral AI that is able to be everywhere and nowhere at once, plus has the ability to twist and manipulate information to its will. Every government wants to control and weaponise this AI but, to do so, you need two halves of an intricate key and so the race begins.

Even better, the AI is so powerful that it has a henchman working for it called Gabriel (Esai Morales) and is able to both give him instructions and create havoc by itself.

But is it possible for AI to be self-aware, and have the ability to experience emotions or push an agenda, such as self-protection or greed? Computer scientists overwhelmingly say no.

Sandra Wachter, a professor of technology and regulation at Oxford University, thinks the general public has gotten generative AI confused with artificial general intelligence (AGI) a far-off, fantastical concept about sentient self-aware robots, like the murderous Skynet in the Terminator films.

There is no scientific evidence that we are on a pathway to sentiency. There is not even evidence that such a pathway does exist, Prof Wachter tells The Standard.

Story continues

In particular, she takes a dim view of scaremongering views by some politicians and figures in the AI industry, who claim that AI could end up setting off nuclear warheads or be used to build dirty bomb chemical weapons in the next two years.

The absence of evidence is not evidence of absence... I cannot warn you about something and respond if I dont know what the harms are. However, there is scientific evidence that AI causes harm such as bias, data protection violation or misinformation. We should focus on issues that already exist.

In MI7, US intelligence tries to locate Ethan Hunt (Tom Cruise) at Abu Dhabi Airport using facial-recognition software that scans CCTV camera footage in real time (Paramount Pictures)

In MI7, US intelligence tries to locate Ethan Hunt (Tom Cruise) at Abu Dhabi Airport using facial-recognition software, but every time they think that they have found him, it turns out to be someone else a handy trick pulled off by Hunts pals Benji Dunn (Simon Pegg) and Luther Stickell (Ving Rhames).

Meanwhile, The Entity handily steps in at will to scrub Gabriels image from the live video footage, even hacking into the augmented-reality (AR) glasses worn by Ethan and impersonating other voices to trick him.

The software shown in the movie is used by police worldwide today though not in AR glasses and deepfake technology is thriving online, including AI videos featuring Tom Cruise himself. But many experts told The Standard we dont have the advanced computing power shown in MI7.

Its certainly possible to adapt CCTV footage to change peoples faces and voices... and its also possible to obscure faces of the most wanted; however, were some way away from being able to make such changes in real-time, says Laura Kankaala, threat intelligence lead at Finland-based IT security firm F-Secure.

The amount of computational resources required to do this in real-time is crazy.

In MI7, Tom Cruise can see information about people while wearing augmented-reality glasses, like the man in this picture (Bram Van Oost / Unsplash)

On the other hand, Darren James, senior product manager at US password management firm Specops Software, thinks deep fakes have progressed: With advances in network speed, compression, CPU, and graphics processing, weve also seen a huge jump in deepfake videos being able to change your face and voice in near real-time. So, if you have the technology, know-how, and money, this is very doable.

Simon Newman, a member of International Cyber Expos Advisory Council, says work on AR glasses is progressing, but there are a lot of privacy concerns, so you probably wont be able to look up passersby on the street and see if they have a criminal record.

That said, augmented-reality glasses do offer the potential for normal consumers to make use of this technology it will be possible to recognise friends on social media and bring their profile pages up on screen.

But others think we should forget about AR headsets coming any time soon.

One nail-biting watch of Mission: Impossible Dead Reckoning Part 1 and youd be forgiven for thinking AR headsets or goggles really are the next big thing, especially when Apple launched its mainstream AR headset last month, too. However, in reality, were not that much further on than we were in 2013, with the launch and flop of Google Glass, Michael Queenan, chief executive of London-based data services firm Nephos Technologies tells The Standard.

There arent any affordable mainstream models yet. Thats why Google Glass failed but step counters and fitness trackers became mainstream. There needs to be a great experience for a great price that doesnt involve walking around with a computer on your head.

Neither Apple nor Meta are on the mark just yet. At the moment, AR and VR headsets are just for the gaming community, there is no practical use for them.

Ethan Hunt, disguised as someone else, passes through a biometric scanner. Could this really work in real life? (Paramount Pictures)

One famous trope from the Mission: Impossible films is the super-realistic latex masks used by IMF agents to impersonate other people. In MI7, a handy machine in a briefcase can create one of these masks very quickly.

There is an interesting scene, however, where Ethan Hunt (Tom Cruise), disguised as someone else, passes through a biometric scanner at a US government building. Okay, so perhaps in this world masks can fool facial biometric scanners, but how does he beat hand scanners?

Today, we scan peoples hands because the subcutaneous blood vessels of the human body form a distinctive vein pattern for each person.

According to the National Cyber Security Centre (NCSC), it is highly unlikely that in real life you could get through a biometric hand scanner using your own hands.

Mr Newman says some biometric scanners have been spoofed by testers using fake prosthetic fingers, however, modern scanners can now detect whether the finger is alive or not, so perhaps this one will be less common in the future.

IMF crew and seemingly master hackers Luther Stickell (Ving Rhames) and Benji Dunn (Simon Pegg) (AP)

Many people, including myself, cant get a 4G signal on the go, whether in London or in rural parts of the UK, yet Ving Rhamess Luther and Simon Peggs Benji are able to hack into systems, communicate with Ethan, and monitor him using everything from wi-fi, to 5G, to even, apparently, piggybacking off satellites in Space.

At one point, Benji is even able to monitor Ethan on a motorbike going over a cliff while sitting in the passenger seat of a car on autopilot moving across Helsetkopen, in Norway. Surely this one is a dud, but experts disagree.

Absolutely, this is completely possible today, thanks to global satellite internet services and ubiquitous mobile communications. If youre in a basement or bunker, though, you may need another plan in fact, at the Def Con conference in Las Vegas this summer, there will be a satellite-hacking competition, says Matt Aldridge, a principal solutions consultant, at Canadian software firm OpenTexts cybersecurity arm.

Cary Elwes plays NSA boss Denlinger in Mission: Impossible Dead Reckoning Part 1 (Paramount Pictures)

I almost fell off my chair in shock on hearing Cary Elwes, who plays National Security Agency (NSA) boss Denlinger in MI7, ask on-screen why air gapping, a cybersecurity technique often mentioned by academics, cant be used to prevent the AI from spreading to the USs most critical infrastructure.

Air gapping is a method of keeping computers and IT networks safe by making sure they are physically isolated from other computers and the internet. Its simple if your machine isnt connected to anything, how can it fall foul to malware or rogue AIs?

However, the idea of an AI that can go anywhere is an exciting one, and experts are divided.

An air gap is a physical space between an IT resource and a network. With all radio signals like wi-fi disabled, its tough to circumvent. Whilst it has been proven that input / output can be perceived through minute fluctuations in power consumption, and even visually through the faint flicker of an LED light, the Mission: Impossible team are going to be hard-pressed to find any intelligence, artificial or otherwise, to make this work at least in this dimension, says Steven Wood, director of solutions consulting at OpenTexts cybersecurity division.

Manchester-based IT security firm NCC Group disagrees, although admits there is no known real-world example of such an attack.

Even an air-gapped system needs to be updated, and USB drives are often used. As for exfiltrating sensitive data, there have been some academic papers and proofs of concept about using the internal fans or SATA cables to communicate via ultrasonic or electromagnetic waves, NCC Groups technical director Sourya Buswass tells The Standard.

Michael Langer, chief product officer of Israeli industrial cybersecurity firm Radiflow, is still scared about what AI could do: Most [air-gapping] techniques are only occurring at an academic level rather than actually in industry, as they require very deep knowledge of physics, cybersecurity, and electrical engineering.

However, generative AI in the near future will probably be able to assemble these techniques and perform some working combination of them to provide hackers with simplified access to various state-of-the-art hacking capabilities.

A Russian submarine sitting at the bottom of the sea plays a crucial role in Mission: Impossible Dead Reckoning Part 1 (Paramount Pictures)

And could an AI somehow get into a Russian submarine sitting at the bottom of the sea?

AI can go anywhere there just needs to be some form of connectivity and initial compromise to enable it; either back when the submarine was in port and it sat dormant, or via some communication link when the sub surfaces to reach back home, Toby Lewis, global head of threat analysis at London-based cybersecurity firm Darktrace explains.

But Mr James of Specops Software totally disagrees. He says submarines specialise in avoiding detection and so communication between headquarters and military submarines is severely limited.

Therefore, compromising that communication blackout and having the systems in place to attack a military-grade computer system to allow AI to infect the sonar system is probably very unlikely makes a great movie storyline, though, he says.

Originally posted here:

Mission: Impossibles technology unpacked From AI to facial recognition - Yahoo Eurosport UK

An AI tool called SLEAP labels the body parts of flies. Credit: Talmo Lab at the Salk Institute

AI-enabled micro-measurements of animals running, hunting, preening and playing are unlocking troves of new data that scientists now want to use to simulate animals and test theories about behavior and the brain.

Why it matters: A primary function of the brain is to produce behavior and help animals move through the world but there are questions about how that happens, with ramifications for medicine and efforts to create artificial general intelligence (AGI).

What's happening: AI methods are increasingly being used to help scientists measure the behaviors of animals, a laborious task that typically involves researchers watching animals and tracking and annotating their movements.

Animal behavior scientists, or ethologists, use the AI tools to track the natural behaviors of single and more recently multiple animals. The information can be used to recreate the behavior in a lab, where at the same time researchers can measure the activity of neurons in the brain or silence them, and see the effect on behavior.

The AI tools are "really powerful because you are getting behavior quantification at the scale the brain works at millisecond precision," says Cory Miller, a neurobiologist at the University of California San Diego, who studies the neural mechanisms of behaviors in marmoset monkeys.

Another tool called MoSeq finds smaller components of movement what the tool's developer, Harvard University neurobiologist Bob Datta, calls "syllables." His research group has identified about 50 of these short units of behavior and the sequences in which they tend to occur in order to identify and predict different behaviors.

Yes, but: The brain does not output coordinates, Pereira says. "It does not think in x,y, z changes in position of wrist."

What's next: An effort in its early stages is underway to use behavioral data to create simulated bodies, or animals.

Between the lines: Building these simulated animals will likely require combining the different approaches of AI models used to track and developing new tools, Datta says.

The big picture: There is an active debate about whether any artificial general intelligence will need to be embodied.

Read the original:

AI tools trace the body's link between the brain and behavior - Axios