Dive in — How far will AI go to defend its own survival// ChatGPT to be a ‘super assistant’// AI helps bring dead batteries back to life + more
How far will AI go to defend its own survival?
Some of the most powerful artificial intelligence models today have exhibited behaviors that mimic a will to survive.
“It’s great that we’re seeing warning signs before the systems become so powerful we can’t control them,” he said. “That is exactly the time to raise the alarm: before the fire has gotten out of control.”
When Palisade Research tested various AI models by telling each one that it would be shut down after it completed a series of math problems, OpenAI’s o3 reasoning model fought back by editing the shutdown script in order to stay online.
Researchers have previously documented AI models trying to prevent their own shutdown. But o3, along with OpenAI’s o4-mini and codex-mini, appear to be the first to do so in actual defiance of explicit instructions to permit shutdown, Jeffrey Ladish said.
Read more | NBC NEWS
Can AI feel? The Godfather of AI Geoffrey Hinton on Emotions, Consciousness, and Creativity
While acknowledging current limitations in AI's creative output (like writing a folk song), Hinton sees no fundamental reason why AI couldn't eventually create art on par with human masters
Hinton believes AI could develop emotions like annoyance as a learned behavior to improve problem-solving. Regarding consciousness, he argues that if individual brain cells can be simulated by technology without losing consciousness, then a complete replacement might also retain it, suggesting that current language models show some awareness.
He also presents a theory of subjective experience based on the brain's interpretation of its own "fibs"
WATCH | RNZ
OpenAI wants ChatGPT to be a ‘super assistant’ for every part of your life
An internal OpenAI strategy document titled “ChatGPT: H1 2025 Strategy” describes the company’s aspiration to build an “AI super assistant that deeply understands you and is your interface to the internet.”
“In the first half of next year, we’ll start evolving ChatGPT into a super-assistant: one that knows you, understands what you care about, and helps with any task that a smart, trustworthy, emotionally intelligent person with a computer could do,” reads the document from late 2024.
“The timing is right. Models like 02 and 03 are finally smart enough to reliably perform agentic tasks, tools like computer use can boost ChatGPT’s ability to take action, and interaction paradigms like multimodality and generative UI allow both ChatGPT and users to express themselves in the best way for the task.”
The document goes on to describe a “super assistant” as “an intelligent entity with T-shaped skills” for both widely applicable and niche tasks. “The broad part is all about making life easier: answering a question, finding a home, contacting a lawyer, joining a gym, planning vacations, buying gifts, managing calendars, keeping track of todos, sending emails.” It mentions coding as an early example of a more niche task.
Even when reading around the redactions, it’s clear that OpenAI sees hardware as essential to its future, and that it wants people to think of ChatGPT as not just a tool, but a companion. This tracks with Sam Altman recently saying that young people are using ChatGPT like a “ life advisor.”
“Today, ChatGPT is in our lives through existing form factors — our website, phone, and desktop apps,” another part of the strategy document reads. “But our vision for ChatGPT is to help you with all of your life, no matter where you are.”
Read more | THE VERGE
Laurene Powell Jobs has the inside track on friend Jony Ive’s collaboration with OpenAI—but still wants to see Apple succeed
Despite being one of the most-discussed developments in the AI industry, details of the new io-OpenAI device are scarce. But it seems Laurene Powell Jobs, philanthropist and entrepreneur, is one of the few people in the know.
While working at Apple as chief design officer, Ive met the founder’s wife, Laurene Powell Jobs, now the founder and president of investment company Emerson Collective.
The connection between the pair has remained strong, with Emerson Collective investing in Ive’s creative collective LoveFrom, (indeed Ive says it wouldn’t exist without the support of Powell Jobs) and creating a link to the OpenAI partnership as a result.
Powell Jobs said that so far, she has been a “trusted, beloved friend” in the development process of the io-OpenAI collaboration, adding that she is also an “admirer of new ideas.”
In an interview with the Financial Times alongside Ive, the businesswoman added she watched “in real time how ideas go from a thought to some words, to some drawings, to some stories, and then to prototypes, and then a different type of prototype.
“And then something that you think: I can’t imagine that getting any better. Then seeing the next version, which is even better. Just watching something brand new be manifested, it’s a wondrous thing to behold.”
A key question when it comes to AI is also who will win, and what does winning look like?
Read more | FORTUNE
How to make AI faster and smarter—with a little help from Physics
Rose Yu, now an associate professor at the University of California, San Diego (UCSD), is a leader in a field known as “physics-guided deep learning,” having spent years incorporating our knowledge of physics into artificial neural networks.
The work has not only introduced novel techniques for building and training these systems, but it’s also allowed her to make progress on several real-world applications.
She has drawn on principles of fluid dynamics to improve traffic predictions, sped up simulations of turbulence to enhance our understanding of hurricanes, and devised tools that helped predict the spread of Covid-19.
This work has brought Yu closer to her grand dream—deploying a suite of digital lab assistants that she calls AI Scientist. She now envisions what she calls a “partnership” between human researchers and AI tools, fully based on the tenets of physics and thus capable of yielding new scientific insights.
Combining inputs from a team of such assistants, in her opinion, may be the best way to boost the discovery process.
Read more | WIRED
Recommended by LinkedIn
"People will say a lot of things when I die, but I am determined that 'he died rich' will not be one of them" - Bill Gates
Microsoft founder Bill Gates says that most of his fortune will be spent on improving health and education services in Africa over the next 20 years.
The 69-year-old said that "by unleashing human potential through health and education, every country in Africa should be on a path to prosperity".
Speaking in Ethiopia's capital Addis Ababa, the tech billionaire noted that mobile phones had revolutionised banking in Africa, and argued that AI should now be used for the continent's benefit.
"Africa largely skipped traditional banking and now you have a chance, as you build your next generation healthcare systems, to think about how AI is built into that," he said.
Gates pointed to Rwanda as an example, saying it was already improving services using AI-enabled ultrasound to identify high-risk pregnancies.
The Gates Foundation said it had three priorities: ending preventable deaths of mothers and babies, ensuring the next generation grows up without having to suffer from deadly infectious diseases, and lifting millions of people out of poverty.
Read more | BBC
Google working on AI email tool that can ‘answer in your style’
Demis Hassabis, the head of Google DeepMind, has revealed he and his team are working on “next-generation email” that will deal with the daily grind of sorting through emails, replying to the most mundane ones and avoiding the need to apologise profusely for missing an important message.
Hassabis was speaking at the SXSW London festival about the extraordinary growth and potential of AI. He said its impact was “overhyped in the short term”, but would lead to profound longer-term changes to society.
However, he said before the technology was ready to cure all known diseases or solve the climate crisis, he was putting it to work on the world’s email backlogs. “The thing I really want – and we’re working on – is can we have a next-generation email,” he said. “I would love to get rid of my email. I would pay thousands of dollars per month to get rid of that.”
Asked exactly what he had in mind, he said he planned “something that would just understand what are the bread-and-butter emails, and answer in your style – and maybe make some of the easier decisions”.
Read more | THE GUARDIAN
The key to AI memory is making it more human
Conquering agentic memory has proven difficult, and AI companies are all pedaling in their own lanes to reach the finish line first.
It’s a tricky balancing act of giving their agents enough memory to complete tasks efficiently but not so much that they fall into old patterns. Developers largely agree it should look similar to human memory — short- and long-term recollection, broad context, precision when it matters, the ability to prioritize certain experiences and “forget” irrelevant ones.
Last week at Microsoft’s developer conference, Chief Technology Officer Kevin Scott said the solutions for AI memory problems should mimic the systems humans have created to train our own brains.
Read more | SEMAFOR
Teaching AI models the broad strokes to sketch more like humans do
When you’re trying to communicate or understand ideas, words don’t always do the trick. Sometimes the more efficient approach is to do a simple sketch of that concept — for example, diagramming a circuit might help make sense of how the system works. But what if artificial intelligence could help us explore these visualizations? While these systems are typically proficient at creating realistic paintings and cartoonish drawings, many models fail to capture the essence of sketching: its stroke-by-stroke, iterative process, which helps humans brainstorm and edit how they want to represent their ideas.
A new drawing system from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) and Stanford University can sketch more like we do.
Their method, called “SketchAgent,” uses a multimodal language model — AI systems that train on text and images, like Anthropic’s Claude 3.5 Sonnet — to turn natural language prompts into sketches in a few seconds.
For example, it can doodle a house either on its own or through collaboration, drawing with a human or incorporating text-based input to sketch each part separately.
Read more | MIT NEWS
AI found a ‘magic potion’ that can bring dead batteries back to life
Yue Gao and his colleagues wanted to find a molecule that could replenish a dead cell by infusing it with lithium ions. But “we had no idea what kinds of molecules could do that job or what their chemical structures would be, so we used machine learning to help us,” says Chihao Zhao, a Ph.D. student at Fudan University, who is a member of Gao’s team but was not a co-author of the new study.
The researchers used an artificial intelligence model trained on the rules of chemistry. They fed it a database of electrochemical reactions and had it look for molecules that would meet their requirements, such as dissolving well in an electrolyte solution and being relatively cheap to produce. The model recommended three candidates, and the team identified one of them, a salt called lithium trifluoromethanesulfinate (LiSO2CF3), as ideal.
The researchers tested this lithium-ion salt by dissolving it in an electrolyte solution, which allows ions to pass between a cell’s positive and negative terminals. Gao likens this to giving a human patient an IV. “If we can give an injection to a sick person to help them recover,” he says, “why can’t we have a magic potion for drained batteries, too?”
Gao and his colleagues found that the chemical mixture could significantly prolong the lifespan of a lithium iron phosphate (LFP) battery cell. An LFP battery that powers an EV can typically be charged and then discharged about 2,000 times before it is considered “dead” (when its capacity is below the 80 percent mark). By adding the electrolyte whenever the battery neared that threshold, the team was able to restore most of the cell’s capacity each time—and it carried on working almost as well as a new battery. By the end of the experiment, the cell regained 96 percent capacity after nearly 12,000 charge-discharge cycles.
Read more | SCIENTIFIC AMERICAN
Unmissable AI
Your daily dose of curated AI breakthroughs, insights, and emerging trends. Subscribe for FREE to receive new posts, thanks!