My wife’s procedure went well, and she is now back to her usual self. Thanks to all who sent their good wishes, and also thanks to everyone for your patience with the delay.

China has a different vision for AI. It may be smarter.

Top AI companies in the US are pouring enormous resources into the pursuit of Artificial General Intelligence (AGI - AI as smart as a human being in all domains of knowledge) and even Artificial Superintelligence ( ASI - AI that is smarter than any human). They may be wasting their time. There is no guarantee when, if ever, AGI will be possible. (AI has already achieved superhuman results in several specific domains, but no one has a roadmap for how to get from where we are to a generally intelligent model.)

China, in contrast, is emphasizing the development of practical, real world solutions in agriculture, policing, manufacturing, and consumer services. This bottom-up approach, rather than chasing the moonshot of AGI, aligns well with China’s goals. It will help them develop a tech infrastructure less reliant on the US, to innovate their way past US chip export restrictions, to reduce the energy demands and environmental costs of the US compute-heavy approach, to focus efforts on domestic open source models rather than Western closed-source systems, and to become an AI leader for developing countries in the Global South who need less costly and more immediately practical AI solutions. China is betting that the tortoise of incremental improvement will, over time, outpace the hare of moonshots.

China has a different vision for AI

Billionaires pour $100 million into PAC to elect an AI-friendly Congress

Silicon Valley billionaires have created a $100 million Political Action Committee to try to sway the upcoming 2026 midterm elections by opposing candidates that want to regulate AI. They say that regulation will just slow down US progress in AI, ceding leadership to China. They are modeling their efforts on the highly successful pro-crypto Fairshake super-PAC that helped secure the election of Donald Trump.

Billionaires pour $100 million into PAC to elect an AI-friendly Congress

Meta continues to flail in AI after monster hiring spree, but Zuck’s new BFF Trump is bailing him out

After a whirlwind hiring spree of AI talent, where top prospects were offered hundreds of millions of dollars to sign, (and one dude turned down an insane $1.5 billion offer), Meta/Facebook’s efforts in AI are still floundering. The company’s AI division has undergone at least 2 major reorganizations in the past month, and several new hires have already left after viewing the internal chaos. And Meta still has no AI model to rival the current crop of both closed-source (ChatGPT, Claude, Gemini) and open source models (DeepSeek, Qwen, etc.).

But not to worry. CEO Zuckerberg’s relationship with Donald Trump has warmed considerably since The Donald called for Zuck to be imprisoned. After a meeting with Trump soon after his election last November, Zuck jettisoned nearly all content moderation on his social media platforms, allowing far-right views free expression without fact-checking. He also contributed $1 million to Trump’s inauguration festivities, and paid $25 million to settle a lawsuit over Meta’s suspension of Trump’s account after the January 6 assault on the Capitol. Now Zuck is counted as a trusted advisor, meeting with Trump and figures from his administration often. In return, Trump is fighting for Meta’s priorities such as combating regulation of AI. Trump has also threatened punitive tariffs against countries that want to impose a tax or fines on US digital platforms. It’s nice to have friends in high places.

Meta continues to flail in AI after monster hiring spree

Zuck’s new BFF Trump keeps bailing him out

Anthropic settles with authors in multi-billion-dollar copyright suit

AI startup Anthropic, creator of the popular Claude chatbot, has settled what could have been a disastrous copyright lawsuit for the company. In June, a federal judge handed down a split decision in a lawsuit brought by authors and copyright holders over Anthropic’s use of copyrighted material gleaned from the internet to train their AI models. Multiple other AI companies are embroiled in similar suits, notably OpenAI’s battle royale with the New York Times. The first part of the judge’s decision in June was highly favorable to Anthropic: the judge ruled that using copyrighted material for training AI chatbots was likely “fair use” since the resulting productions of the chatbot were transformations of the original work, not just copying. Unfortunately for Anthropic, though, they had knowingly used a repository of 7 million pirated works in their training, and the judge saw that as a clear violation. The fine for using a pirated work can be as high as $150,000 each; for the 7 million works used, the fine could total to a staggering $1.05 trillion. Even for an AI company with lofty valuations, this would be a corporate death sentence. As the saying goes, once you have a firm grasp on a man’s tender parts, his heart and mind will usually follow. Anthropic came back to the negotiating table with an open checkbook, and the case will be settled for a tiny fraction of the potential trillion-dollar liability. (No public word yet on the final price.)

Anthropic settles with authors in multi-billion-dollar copyright suit

Study finds that quiz on faked images increases trust in news sources

Researchers working with a major German newspaper (Suddeutch Zeitung) presented visitors to the paper’s website with a quiz asking whether the reader could tell the faked AI-produced image from a real photograph for 3 image pairs. Only 2% of test takers got all 3 answers right, and 36% got all 3 wrong. Follow up showed that test takers visited the paper’s website more frequently after the quiz, and subscribers were less likely to let their subscription lapse. The authors concluded that being confronted with credible faked information made a user value trusted news sources more. This presents a ray of hope that as our information spaces get ever more polluted with misinformation and disinformation, users will increasingly turn to news sources that they trust. We can only hope.

Quiz on faked images increases trust in news sources

Ethan Mollick’s essay on Mass Intelligence

Ethan Mollick, a professor at the Wharton School of Business, is one of the leading thinkers on the business and social implications of AI. His latest essay notes that AI is getting increasingly powerful, increasingly accessible, and increasingly easy to use. He notes that today more than a billion people use one or another of the top AI models (ChatGPT, Claude, Gemini, etc.) regularly. He argues that we are transitioning from the Information Age to the Age of Mass Intelligence. And none of us - literally no one - is prepared for what that means.

Here, in his own words, is the crux of his essay:

“Every institution we have — schools, hospitals, courts, companies, governments — was built for a world where intelligence was scarce and expensive. Now every profession, every institution, every community has to figure out how to thrive with Mass Intelligence. How do we harness a billion people using AI while managing the chaos that comes with it? How do we rebuild trust when anyone can fabricate anything? How do we preserve what's valuable about human expertise while democratizing access to knowledge?”

Ethan Mollick essay on Mass Intelligence

AI vision plus “functional inks” track product safety and freshness

Functional inks are inks that change color, sometimes very subtly, in response to environmental variables such as heat and humidity. Foods or electronics that are exposed to excessive heat or moisture can spoil or become damaged. Traditional “smart packaging” has used electronic sensors to track these variables, a solution that is effective but too costly for many applications. Recent research has shown that AI vision systems can track subtle changes in functional inks printed on the outside of a package, giving a cheaper and finer-grained view of the safety and fitness for use of the products inside.  In these cases, the packaging is the sensor.

AI vision plus “functional inks” track product safety and freshness

Fighter pilots take battle directions from AI in Pentagon test

In a recent joint military exercise between Air Force and Navy, fighter pilots for the first time tested an AI “air battle manager.” In modern air combat, fighter pilots typically take direction from human battle managers, who monitor radar, sensor feeds, and and intelligence to direct pilots where to fly and how to position their aircraft. In this test, orchestrated by the Pentagon, fighter pilots for the first time took direction from Raft AI’s Starsage tactical control system. The major advantage of the system is speed, with decisions made in seconds rather than the minutes required in a human-controlled system.

Fighter pilots take direction from AI in Pentagon test

.

AES deploys fleet of AI-enabled robots to install solar panels

Global energy company AES is currently deploying a fleet of AI-enabled robots to install solar panels in California. There they will assist with utility-scale installations, which are becoming ever more common The company is at pains to mention that the robots integrate “seamlessly” with union labor.

AES deploys fleet of AI-enabled robots to install solar panels

AI-powered dolls are becoming caregivers in South Korea

South Korea has one the most rapidly aging populations in the world. Young Korean women, highly educated but caught in a traditional culture that expects women to make most of the sacrifices for family, are increasingly declining to have children, or even to marry. And Korean culture is hostile to immigration. The result is that each succeeding generation is less numerous than the last. This means that there are not enough younger caretakers to care for the growing population of elderly. One solution being explored is using robots for caregivers. Korean startup Hyodol markets cute AI-powered dolls that keep elders company and monitors their health and wellbeing. The doll will chat with the elder if the individual desires, will remind them to take their medicine, monitors their physical activity, and can send a voice alert to family or staff for any worrisome events. The company emphasizes emotional support and dementia prevention as core benefits of their AI-powered doll.

AI-powered dolls are becoming caregivers in South Korea

UK doctors develop an AI stethoscope

The venerable stethoscope was invented almost by accident in 1816, when a shy French physician searched for a way to listen to the heart of a very buxom young woman without pressing his ear to her ample chest, as was the common practice of the day. In a moment of desperate innovation, he rolled up a sheaf of manuscript paper and found that the air tube conducted sound splendidly. Et voila! the stethoscope was born.

The design of the stethoscope has changed remarkably little in the intervening 200 years. Until now. Now researchers at Imperial College London have developed an AI-enabled stethoscope that can diagnose heart failure, valvular disease, and abnormal heart rhythms almost instantly. The device, about the size of a deck of cards, takes an EKG while recording sounds of the patient’s heart. The EKG and sound data are analyzed by remote AI models through a link to an app on the physician’s smartphone. Within 15 seconds, the results of the analysis appear in the app. A review of the use of this device in 12,000 patients found that occult heart disease was diagnosed significantly more often with the device than with a standard stethoscope. EKO, the company that developed this device, is already marketing various versions of it to physicians in the US and UK.

UK doctors develop an AI stethoscope

(Answer to AI faked photo quiz: the photo on the right is AI generated.)