Nobelist Demis Hassabis says key skill in the Age of AI is “learning how to learn”

Like a modern-day Oracle of Delphi, Nobel Prize-winning AI scientist Demis Hassabis, head of Google’s DeepMind AI research lab, sat with Prime Minister of Greece Kyriakos Mitsotakis in an ancient outdoor theater in Athens before an audience of 5,000 attendees, and made pronouncements on humanity’s future in the coming Age of AI. He warned that the pace of change is so rapid that “the only thing that you can say for certain is that huge change is coming.” He scoffed at the idea that AGI (AI equal to humans at all tasks) was near, but allowed that it was possible in 10 years. At that point, science and technology would progress so swiftly that a future of “radical abundance” was possible. Coping with this accelerated progress will require lifelong learning, including learning how to learn more quickly and effectively than traditional methods are capable of.

Nobelist Demis Hassabis says key skill in the Age of AI is “learning how to learn”

OpenAI and Anthropic reveal how people are actually using AI

Last week both OpenAI and Anthropic (creator of the Claude chatbot) released major studies on how their users actually use AI. 

OpenAI analyzed usage of their consumer accounts, which are held by individuals, as opposed to their business accounts. They found that over the past 2 ½ years, individuals are using their personal accounts less and less for work, and more and for personal uses such as Practical Guidance (getting advice on workout routines, using ChatGPT as a thought partner to generate ideas or critique a decision, etc.), Seeking Information (replacing traditional search), and Writing (2/3 of these requests are for suggestions on editing or summarizing something already written.) In that period, the gender gap, in which the majority of early users were male, has largely disappeared, with near-equal participation by males and females.

Anthropic analyzed both their consumer accounts and their business accounts. They noted that AI adoption by employees at work in the US has doubled in the past 2 years, from 20% to 40%. This rise is faster than any previous technology wave, including personal computers, the internet, and smartphones. Even so the adoption of AI is geographically concentrated, with more prosperous locales having AI usage many times that of less prosperous ones. For example, Israel has an AI Usage Index that is 25 times that of India. Business use is divided into automation vs. collaboration. Less prosperous locales, as well as all businesses, focus on automation. Business use by individuals in prosperous locales tends to favor collaboration, with AI as a copilot or thought partner.

How personal account users use ChatGPT

How personal and business account users use Claude

Microsoft and OpenAI make it official: No longer exclusive, but still good friends (or so they say)

Once the closest of partnerships, with Microsoft investing an (at the time) staggering $13 billion in OpenAI in return for use of its leading-edge AI technology, the relationship between the two has steadily devolved into more of a situationship. Microsoft wants OpenAI to help keep it at the top of the tech world; OpenAI wants the freedom to ascend to the top itself. Microsoft holds a veto over OpenAI’s transition from a nonprofit to a public company, which is the key to OpenAI’s unlimited future. After both sides played footsie with other vendors (Microsoft added Anthropic to its AI lineup; OpenAI made a historic deal with Oracle for cloud services), the two tech giants have finally released a joint public statement on the state of their relationship.

Tl;dr -it’s complicated. Microsoft will still get access to OpenAI’s technology, and will allow OpenAI to restructure itself and use other cloud vendors. Beyond that - it’s murky. But at least for now, both sides seem to feel that hey are getting enough of what they need to move forward amicably.

OpenAI has other problems with the restructuring. Attorneys General in both California and Delaware need to sign off on the conversion to for-profit. AGs tend to be skeptical of such deals unless 100% of the value of the for-profit arm is put into the nonprofit - essentially buying the nonprofit out at full market value. OpenAI is offering far less than that, but still a substantial $100 billion, which would make it one of the largest nonprofit foundations in the world.

Elon Musk has vowed to fight the restructuring in court, so expect more drama.

Axios’ explainer of the complex OpenAI deals in play

Oracle stock soars on news of $300 billion OpenAI deal; CEO Ellison is (briefly) world’s richest man

OpenAI’s insatiable need for cloud computing resources to fuel its massive AI business (700+ million weekly users and climbing) just got a new source - it has inked a deal with Oracle Corporation for $300 billion in cloud computing over 5 years. One obvious flaw in this historic deal - Oracle doesn’t (yet) have the computer capacity, and OpenAI doesn’t (yet) have the money. No matter. Oracle’s stock soared by 40% in one day after the news broke, making Ellison richer than Elon Musk for a few hours, until the euphoria ebbed a bit. (At nearly the same time, Elon hyped Tesla stock by using $1 billion of his own money to buy shares, which excited investors who had been discouraged by Tesla’s declining sales, revenue, and profit amid widespread anti-DOGE sentiment and fierce competition from Chinese electric vehicle manufacturer BYD.)

Oracle stock soars on news of $300 billion OpenAI deal

Albania appoints AI as minister of government

Albanian government procurement has long been riddled with corruption. To address this problem, Alabanian Prime Minister Edi Rama announced last week that he was appointing “Diella”, the nation’s AI-powered virtual assistant for government services, which helps citizens obtain drivers licenses and pension applications, as the new Minister overseeing all procurement from private companies. The thought seems to be that a machine intelligence will be incorruptible because it has no use for money or favors. The problem is that all computer systems are potentially hackable, so the machine’s decisions will have to be overseen by - fallible, corruptible - people. Time will tell whether this was merely a publicity stunt, or just a really naive idea.

Albania appoints AI as minister of government

Vibe coding is turning senior software developers into “AI babysitters”

AI has gotten so good at writing software that many senior software developers are turning to it to automate some of the more tedious aspects of their job, such as putting in boilerplate code for basic functionality. This practice of using text prompts to ask AI chatbots to produce computer code has been termed “vibe coding.” As the coding chatbots rapidly improve, many senior developers are beginning to see their job transition from directly writing code to overseeing the code that the chatbot produces, or as some have called it, “babysitting the AI.” Some developers lament the change, but many see the tradeoff as worth it, because they are able to produce substantially more lines of code per day, sometimes as much as 2 ½ times the amount they can produce manually. This transition from directly performing a task, to supervising AI performing the task, is likely to affect more and more knowledge workers over time.

Vibe coding is turning senior software developers into “AI babysitters”

Google Gemini tops App Store downloads, supplanting ChatGPT

ChatGPT has long been the most downloaded free app in the Apple App Store, but the frenzy over Google’s viral hit, the Nano Banana image editing feature of its Gemini chatbot, has now propelled Gemini to the top spot. Nano Banana’s intuitive text-to-image editing has taken the internet by storm, allowing ordinary users to make sophisticated edits to photos, such as removing background people, changing a person’s outfit, or even replacing the drab original location with a mountaintop view or the surface of the moon. Now every user that wished they could perform sophisticated image editing but found Photoshop too complicated can just tell Gemini’s Nano Banana what they want, and voila! - image editing magic.

Google Gemini tops App Store downloads, supplanting ChatGPT

Malawi farmers use AI to survive climate change

Malawi is a poor, landlocked nation in Eastern Africa where 80% of the country’s economy depends on agriculture, and the vast majority of its people are small subsistence farmers. Devastating cyclones and drought, fueled by climate change, have pushed many of these small farmers to the brink by destroying crops and washing away topsoil. Now their government is trying to use AI to help.

Government agricultural agents are assigned 150 - 200 farms in a local area, and meet with the farmers in groups once a week. The government agents are each issued a smartphone with an account with an AI chatbot designed by the non-profit Opportunity International for localized farming advice. Advice such as switching from corn to potatoes in order to maximize yield from degraded soil has helped many a farmer stave off starvation.

Malawi farmers use AI to survive climate change

Amazon’s Zoox robotaxi begins giving free rides on the Vegas Strip

Amazon-backed Zoox robotaxi company has begun offering users free rides along the Las Vegas Strip. The rides are free because the company does not yet have all the regulatory approvals to offer rides commercially. With 40 million tourists a year, the Strip is a plum spot to publicize the company’s new driverless transportation service.

Unlike rival robotaxi services such as Google’s Waymo which use standard automobiles retrofitted with sensors and autonomous vehicle software, Zoox uses custom-built vehicles with a unique design. Each Zoox taxi is equipped with multiple types of sensors and a unique club car interior where 4 passengers can face each other, and there is no driver’s seat or steering wheel. Amazon is betting that its design, purpose-built for its intended use, will eventually win out over the retrofits.

Amazon’s Zoox robotaxi begins giving free rides on the Vegas Strip

Nursing robots address aging population and nursing shortage

All developed countries have aging populations, due to better access to food and medical care making people live longer, plus a falling fertility rate caused by people reproducing at later ages and choosing fewer - or no - children. This is causing a worldwide nursing shortage. The World Health Organization (WHO) estimates that the global shortfall could rise to 4.5 million nurses needed by 2030.

Technology companies are rushing to fill the gap with robots. Foxconn, the Taiwanese company that makes most of Apple’s iPhones, has entered the race with Nurabot, an autonomous AI-powered semi-humanoid wheeled robot designed to deliver medications, show patients around the ward, and perform other repetitive or physically demanding tasks for nurses. Multiple other companies are pursuing the same market, with widely different designs, capabilities, and price points.

Nursing robots address aging population and nursing shortage

Harvard Medical School has developed an AI system called PDGrapher that pinpoints multiple drivers of each disease, and then suggests drugs or drug combinations that might restore diseased cells to health. This differs from traditional drug discovery, in which one cellular protein is targeted, and researchers try to find a drug that can bind to it in a therapeutic way. Since many diseases are complex, with multiple causes and effects, PDGrapher’s more comprehensive approach can shortcut the time it takes to discover effective treatments. PDGrapher is available for free in a publicly available GitHub repository.

Harvard AI tool pinpoints genes and drug combinations to restore cells to health

Sam Altman-backed Retro Biosciences seeks to extend your life

Sam Altman, CEO of OpenAI, has been making big bets on technology for 20 years. In fact, the majority of his estimated 2.2 billion net worth comes from investments outside of OpenAI.

One major bet is the $180 million he personally invested in longevity startup Retro Biosciences. Retro Bio’s mission is to reverse aging and cure aging-related diseases. It is pursuing several lines of research to achieve these goals, including research on stem cells, autophagy (a cellular clean-up process), tissue re-engineering, and more. They are on track to begin the company’s first clinical trials later this year, on a drug that is designed to cure Alzheimer’s by increasing autophagy in the brain. More drugs are in the pipeline, and they all revolve around trying to move cells to younger versions of themselves, with a goal of keeping patients healthy and functional for at least an additional 10 years.

 Retro Bio wants to extend your life and fix your brain