AI News - November 2024

A quick overview of interesting AI News for the Yate and Sodbury U3A's monthly AI News meeting held on November 14th 2024 (meetings are second Thursday of month, 1400-1545, St John's Church Centre, Wickwar Road, Chipping Sodbury - all welcome). Note that this is very much a summary to act as background as the topics are discussed in detail at the meeting itself. We also discuss and demonstrate other AI products and services and much more.

[1] Tesla's 'We, Robot' Day - the Optimus Robot

The 'We, Robot' Day was held on 10th October 2024 and was about both the launch of the self-driving 'robotaxi' and a reminder about the rapid progress being made with humanoid robots. First we look at the humanoid robots.

[1a] "The Optimus robots will walk among you." They danced, wandered around and spoke to the people in the crowd, and served drinks and snacks. Some aspects were tele-operated (see also 1b. below) but nevertheless it was an impressive display of confidence. Elon says: "The biggest product ever, of any kind".

pic.twitter.com/VK9vlGF0Ms

— Elon Musk (@elonmusk) October 11, 2024

[1b] Optimus posted: "Navigating by myself". Issued one week after the event in response to suggestions that Optimus always needed 'tele-operation'. Notice the stairs and the ability to couple themselves up to the charging station.

Navigating by myself pic.twitter.com/CeFSqCcy5I

— Tesla Optimus (@Tesla_Optimus) October 17, 2024

[2] Tesla's 'We, Robot' Day - the Robotaxi/Cybercab

Tesla's Full Self Driving technology continues to improve rapidly. Many users report long and complex journeys in their Tesla with no need for their intervention. The aim is, of course, to achieve a completely self driving vehicle (no internal steering wheel or brakes in the 'taxi' version).

[2a] First view of the Robotaxi. The event showed 19 Cybercabs and 29 Model Ys driving themselves on 1,300 trips transporting over 2,000 guests

Rides in the city of the future https://t.co/swCCGwLP1X pic.twitter.com/mu5IfQD6AN

— Tesla (@Tesla) October 17, 2024

[2b] How do Robotaxis charge themselves? No need for human or humanoid robot help! Apparently this has an efficiency of approx. 93% and is very similar to wired systems in that regard.

Robotaxi wireless charging

No hands required pic.twitter.com/XL746DkGhb

— Tesla (@Tesla) October 18, 2024

[2c] And the self-driving Robovan has unlimited uses. Likely to be outfitted in a wide range of styles by third party suppliers, so very mobile home, or party bus, or ... insert your use here!

Robovan making an appearance— you can hear me say “Art Deco Style” & “Steamliner” pic.twitter.com/zZ1F3M5g6k

— K10.13✨ (@Kristennetten) October 21, 2024

[3] More on Google's NotebookLM

There are many AI products that will analyse pdfs and other source documents for you. Few have established themselves in the market. NotebookLM added two hosts talking about the subject matter, and that caught the public imagination in a big way.

[3a] Example of NotebookLM I gave NotebookLM the just issued UK Parliamentary Bill on Assisted Dying. It produced a thorough 16 minute podcast.

I put the new Terminally Ill Adults (End of Life) Bill from @kimleadbeater into Google's NotebookLM AI & asked for a podcast. How do you think it did? It is 16 mins long. #AssistedDying #AssistedDyingBill pic.twitter.com/VNJeeDydaN

— The Silver AI Project - free AI training materials (@SilverAIProject) November 12, 2024

[3b] Discussion about the the creation of NotebookLM. Like ChatGPT when first released, this was meant to just be a preview, but had immediate and unexpected success.

NotebookLM from the words of the creators.

This is going to change the world of podcasting... pic.twitter.com/gxMUoEofFC

— Wes Roth (@WesRothMoney) October 17, 2024

[4] Is AI Slowing Down or Speeding Up?

Journalists love to write the story that AI is slowing down and will turn out to have been a bubble. What does the evidence show?

[4a] How ChatGPT o1 learned to think for longer. Noam Brown of OpenAI explains how they discovered a new method. Previously the models just delivered an answer 'in one go' but now 'thinking for longer' gave emergent properties. 'We can push this further'.

Anyone who still believes that the development of AI has reached its limits should watch this video excerpt from Noam Brown (OpenAI).

Sometimes I get the feeling people arent listening closely.

Noam Brown: "We saw that like yeah it, once it's able to think for longer it um… pic.twitter.com/cwqFyX8Eto

— Chubby♨️ (@kimmonismus) November 12, 2024

[4b] Sam Altman of OpenAI talks Artificial Superintelligence (ASI). With a timeline of six or nine years away, this is intelligence capable of discovering all physics, fixing the climate, abundant energy etc etc. 'We are nowhere near the limit of progress'.

Sam Altman says Artificial Superintelligence is thousands of days away. There is a lot of compounding left to do.

What happens then? Fix climate, space colonies, limitless intelligence, understanding physics, abundant energy.

Think bigger. The age of abundance is nigh. pic.twitter.com/i8I3w9hmLl

— Garry Tan (@garrytan) November 8, 2024

[4c] Sarah Friar of OpenAI talks of PhD level AIs. Talks about their 'reasoning models' and how 'It would blow your mind to see what is already coming'. That is, they have some amazing things internally that are not yet released.

OpenAI CFO Sarah Friar says AGI is "closer than most think" and the ability of internal research models to perform at PhD level in a range of fields "would blow your mind to see what's coming" pic.twitter.com/paA8meE7NF

— Tsarathustra (@tsarnick) October 28, 2024

[4d] David Shapiro on when AI has real life effects. '2024 was Year 0, the real beginning of the wave (though many will remember ChatGPT as the first shot across the bow).

2025 is Year 1. That's the year that AI became commercially viable across most sectors, so we'll see a surge of early adopters next year (we're currently in the "innovators" part of the curve now). We'll see some huge strides in commercial humanoid robotics next year, along with everyone agreeing "yeah, okay we have AGI for sure" though there will always be knuckledraggers moving the goalposts.

It'll be like drinking from the firehose more than it is today, but 2026 will be the year of Scorched Earth. What I mean by that is that companies are gonna start imploding, those that fail to adapt. Think Borders. But also jobs attrition will likely be a hot topic ...'

[5] Is the UK ready for rapid AI progress

A recent interview with Keir Starmer suggests not ... Of course, this may be a deep fake. We should hope that it is!

Please tell me Keir didn’t call Google “googoloo” then go on to compare AI to a fork.

“Ai is like a fork, if you’re trying to eat your dinner it’s a useful thing to have… but it can always be used to stab someone”

🤦‍♂️🤦‍♂️🤦‍♂️ pic.twitter.com/uwSClmKCgi

— Alex Armstrong (@alexharmstrong) November 10, 2024

[6] The resistance movement to AI builds

[6a] Anthropic (Claude) danger level charts. Very clever humans rarely also want to cause great harm. However, a very clever AI that is available to all humans poses a far bigger risk.

They've created a system to track the danger:

"ASL levels" (AI Safety Levels) from 1-5

• Current models: ASL-2
• Next year: ASL-3
• 2026: Likely ASL-4

But ASL-3 is the real turning point: pic.twitter.com/B47ujWn6aP

— Ben Averbook (@benaverbook) November 12, 2024

[6b] Many leading figures in AI are concerned. As of yet, the resistance to AI is largely academic with no large 'real world' protest. See this paper for a typical analysis.

As we speed towards AGI and then ASI many think this is a fatal mistake. Here is an example 'Guide to Extinction Risk from AGI'. Suggests that humanity will not survive the emergence of 'God-like AI'.

RELEASE: THE COMPENDIUM

Several reckless groups are racing towards AGI, and risking the extinction of humanity.

What is AGI? Who are these people? Why are they doing this? And what can we do about it?

We answer these questions in The Compendium.

1/17 pic.twitter.com/EiuC5u0fK8

— Connor Leahy (@NPCollapse) October 31, 2024

[7] The benefits of AI

The evidence mounts of how AI already helps in many areas, with examples here from education, commerce and research and development.

[7a] Ethan Mollick on AI Tutors. It takes a long time for proper trials to be run, but Harvard now has shown that personal AI tutors have many benefits.

More evidence that a well prompted LLM can help learning from a randomized controlled trial at Harvard: “here we show that students learn more than twice as much in less time with an AI tutor compared to an active learning classroom, while also being more engaged and motivated.” pic.twitter.com/WChkpHhBG6

— Ethan Mollick (@emollick) October 29, 2024

[7b] Amazon delivery drivers getting AI glasses. Report that human delivery drivers will get smart glasses (before being replaced by humanoid robots no doubt).

Amazon is developing smart eyeglasses for its delivery drivers to guide them to, around and within buildings, as it tries to smooth the final stretch of an order's journey to a customer's home, five people familiar with the matter said.

The glasses would provide drivers with… pic.twitter.com/3oBWysXzyS

— Brian Roemmele (@BrianRoemmele) November 11, 2024

[7c] AI Assisted Researchers benefit. Evidence that AI can contribute in idea generation in material science but most benefits come when working with the most talented humans.

This is the best paper written so far about the impact of AI on scientific discovery pic.twitter.com/OxjsBetcHP

— Caleb Watney (@calebwatney) November 8, 2024

[7d] New Google Learning Tool - Learnabout.

Alongside NotebookLM, Learnabout from Google looks like the next big blockbuster. A tool that virtually gives you a learning unit with instructions on topics on request! Great.pic.twitter.com/U9wgdWkTMX

— Chubby♨️ (@kimmonismus) October 31, 2024

[7e] Robots as doctors.

BREAKING: New frontier in medical robotics 🤖 Trained robot performs with skill equal to a human doctor 🚀👀 The disruption of labor is underway... https://t.co/QQHU7tsWip

Uncover 15 essential insights into the disruption of labor by humanoid robots: https://t.co/8dpznaB2nN

— RethinkX (@rethink_x) November 13, 2024

[8] Robots come in many forms

The rate of development in China in particular is going to stun the world once these many designs reach the market. Whether the AI intelligence aspects are as strong as those from the West, we don't yet know.

[8a] Robotic dogs do more than bite. Future law enforcement and warfare will make full use of all forms of robotic device.

A squad of robotic dogs deploys from a vehicle in China, some appear to be armed.pic.twitter.com/8NPsBjnNb7

— Yam Peleg (@Yampeleg) November 10, 2024

... or see this one ...

Extreme Off-Road | #DEEPRobotics #Lynx All-Terrian #Robot#robotdog #tech #ai #quadrupedrobot #robotics pic.twitter.com/hxoTrvu86c

— DEEP Robotics (@DeepRobotics_CN) November 13, 2024

[8b] Humanoid robots can be small and cute. This is a strange video however it shows the robot walking around, shaking hands and then sitting down to listen to music.

This is probably the best video I've seen in a while about robots. it looks like the Unitree G1 model. Clean and simple, doesn't seem scripted. Wait for the moment it responds to the music and sits down to listen.
China is accelerating on next level. pic.twitter.com/TvHMqUbtG2

— AshutoshShrivastava (@ai_for_success) November 11, 2024

[9] Robot and AI Rights

It has been discussed for many years but we have now reached the point where we have to consider whether AI has welfare needs. After all, if it is smarter than us, could it suffer stress and other forms of pain and anxiety?

SCOOP: Anthropic has hired an AI welfare researcher.

Kyle Fish joined the company last month to explore whether we might have moral obligations to AI systems. pic.twitter.com/1jq8KINSrn

— Shakeel (@ShakeelHashim) October 31, 2024

[10] AI reaches 'Stage 3' - AI Agents

The stages often used by OpenAI are: Stage 1 - Chat, Stage 2 - Reasoning, Stage 3 - AI Agents, Stage 4 - AI Innovators, Stage 5 - Corporate AI. AI Agents can do things in the real world, such as access your computer, send emails, buy and pay for goods and services etc.

🚨Anthropic just released the most amazing AI technology I've ever used

I'm not kidding

AI agents are here and you can now build your own personal army of AI's that will do work for you

Here is your demo and complete beginner's guide:

(trust me, you want to bookmark this) pic.twitter.com/MueqisKpmd

— Alex Finn (@AlexFinnX) October 22, 2024

[11] AI Music improvement

Suno Music v4 suno.com is receiving great reviews, suggesting it approaches human standard. You decide.

When I first got access to @sunomusic v4, the very first thing I did was remaster the best song I ever created with it: "After the Winter" by the fictional artist Steve Rose.

Here it is in gorgeous remastered glory with the original music video. Still an absolute banger of a… pic.twitter.com/Avb2Uv6866

— ⭕ AI & Design (Marco) (@AIandDesign) November 12, 2024

[12] The USA Election Effect

AI is already accelerating large numbers of technologies. The outcome of the USA election will add huge deregulation in the USA, where many companies are based. Together the effect will be staggering.

Upcoming deregulation wave will create an economic sonic boom.

— Naval (@naval) November 6, 2024

and not investment advice ...

Morgan Stanley gives Tesla's Optimus humanoid robot opportunity an implied present value of $485 per share. $TSLA pic.twitter.com/jHPPcYkO3X

— Sawyer Merritt (@SawyerMerritt) November 12, 2024

And in conclusion:

I feel like this may prove to be the most prescient cartoon of the 21st century. pic.twitter.com/OdKdEBsXnC

— Michael Hartl (@mhartl) October 29, 2024

... all of the above thanks to something like this ... (which was apparently made by brilliant.org which is worth a look.)

Amazing graphical representation of a neural net, never seen anything like it. pic.twitter.com/1xUQHAiq50

— Gabriel Elbling (@gabeElbling) October 26, 2024

... and how we really 'grow' neural networks ...

Now imagine if our chips themselves were bio-inspired in their self-organizing behavior.

That's what we're building @Extropic_AI. The future is going to get weird.
pic.twitter.com/YkSp7pEuYL

— Beff – e/acc (@BasedBeffJezos) November 13, 2024