2026 aug 17

[rcjordan]

US to tell partners they must pick sides in AI race with China | Reuters
https://www.reuters.com/world/china/us-tell-partners-they-must-pick-sides-ai-race-with-china-2026-08-14/

Albertsons Says Its Shoppers Are Buying More—Thanks to AI - WSJ
https://www.wsj.com/cio-journal/albertsons-says-its-shoppers-are-buying-morethanks-to-ai-81793975?mod=rss_Technology

Alibaba challenges Meta with new laptop-ready AI model
https://www.cnbc.com/2026/08/17/alibaba-meta-qwen-open-weight-ai-laptop-models.html

The Hottest AI Models Aren't the Ones Developers Actually Use - Business Insider
https://www.businessinsider.com/top-ai-models-usage-data-hugging-face-2026-8

Your AI is emailing my AI—and nobody's in charge
Half of U.S. workers use AI to write their messages.  - Fast Company
https://www.fastcompany.com/91588570/ai-emailing-ai-nobodys-in-charge?partner=rss

Code fixers have fired up the AI warp drive. Strange new worlds await
https://www.theregister.com/columnists/2026/08/17/code-fixers-have-fired-up-the-ai-warp-drive-strange-new-worlds-await/5287681

What AI chatbots can do with your personal data
https://www.axios.com/2026/08/17/secrets-share-ai-openai-meta-google-anthropic

'I see the incredible promise': on set of an AI film shoot as new studios embrace controversial tech | Film industry | The Guardian
https://www.theguardian.com/film/2026/aug/16/directors-embracing-ai-film-making

Why people aren't buying Mark Zuckerberg's AI future | TechCrunch
https://techcrunch.com/2026/08/16/why-people-arent-buying-mark-zuckerbergs-ai-future/

Dario Amodei Says AI Needs to 'Actually' Cure Cancer to Win Over Public - Business Insider
https://www.businessinsider.com/anthropic-ceo-dario-amodei-ai-public-opinion-cure-cancer-2026-8

[Travoli]

What AI chatbots can do with your personal data
>Imagine a chatbot built to keep you talking a little longer. To change your mind about something. Or persuade you to buy something. Now imagine it could draw on everything it knows about you — including the fears, insecurities and private details you shared in conversation — to do it.
>"The more a system knows about you, the easier it will be to make escalating requests for private details in a way that feels natural"

If you thought they knew everything about us already...

Dario Amodei Says AI Needs to 'Actually' Cure Cancer to Win Over Public

That would be a great day.

[rcjordan]

► Climate scientists are not alarmists but have underestimated recent climate changes.
► We identify a directional bias toward erring on the side of least drama (ESLD).
► ESLD is an internal pressure arising from norms of objectivity, restraint, etc.
► ESLD may cause scientists to underpredict or downplay future climate changes.

Climate change prediction: Erring on the side of least drama? - ScienceDirect
https://www.sciencedirect.com/science/article/abs/pii/S0959378012001215

[rcjordan]

Chinese scientists achieve record-breaking 24.0% efficieny for large-area perovskite solar modules - pv magazine Global
https://www.pv-magazine.com/2026/08/17/chinese-scientists-achieve-record-breaking-24-0-efficieny-for-meter-scale-perovskite-solar-modules/

Section 232 minimum import prices force foreign solar to match U.S. production costs - pv magazine Global
https://www.pv-magazine.com/2026/08/17/section-232-minimum-import-prices-force-foreign-solar-to-match-u-s-production-costs/

[rcjordan]

AAAaaaccckkk!

Reddit's AI is turning posts into podcasts and short videos | The Verge
https://www.theverge.com/tech/981289/reddit-ai-text-video-posts

[rcjordan]

Some, but not all....

Be a hater all you want, AI's here to stay
https://www.theregister.com/columnists/2026/08/17/be-a-hater-all-you-want-ais-here-to-stay/5288275

Anthropic's annualized revenue surges to $65B | TechCrunch
https://techcrunch.com/2026/08/17/anthropics-annualized-revenue-surges-to-65b/

[ergophobe]

> cancer

According to Derya Unutmaz in this podcast interview
https://www.foundmyfitness.com/episodes/derya-unutmaz

some guy sequenced the genome of his dog from healthy tissues and a tumor, then had one of the AI models analyze the two and suggest a molecule (a protein I guess?) that he could target, then ordered an mRNA vaccine that targeted the molecule, injected the dog, cured the tumor.

Unverified.


>> including the fears, insecurities and private details

https://en.wikipedia.org/wiki/Ministries_in_Nineteen_Eighty-Four#Room_101

>> ELSD

Maybe so. I think that's true. People want to sound "reasonable." There is another factor that has resulted in early predictions being off: up until about 1990, all the models assumed that once the science was clear, governments would act. But after James Hansen's 1988 testimony before Congress, the Koch brothers and some others realized that the science WAS clear and they poured huge amounts of money into casting doubt on it, which led to a lot of delay in taking action.

In other words, the models were reasonably close on the question of how temperature would rise for a given amount of carbon, but they made rosy predictions about how alarmed people would get.

[ergophobe]

> here to stay

"First, AI has been destroying jobs and will continue to do so. Research house Forrester, for example, predicts that by 2030 AI will wipe out 10.4 million US jobs."

There is a difference between an event that HAS happened and a prediction about what WILL happen.

"important to distinguish AI-generated content from human-created material, yet 53 percent lacked confidence that they could reliably do so"

I think we can say that the 53% who lack confidence that they can reliably distinguish AI content are the 53% who actually use AI and know what AI is producing. The 47% who are confident are the ones who have no idea what is and what isn't AI-generated.

The 47% likely also includes people who used ChatGPT in 2024, decided it hallucinates all the time, and have never used it again, but insist that the output is rife with hallucinations.

QUESTIONS: Can you reliably distinguish AI content from human content? Do you feel that the current frontier models still hallucinate a lot?

I would say definitely no to both.

Of course, LLMs do sometimes hallucinate, but this is more and more like the Waymo Problem. Waymo is way safer than humans, but many people will not get into a car without a driver. Similarly, humans "hallucinate" all the time* in conversation, but people freak out if in a casual chat with an LLM it mixes up some facts.

*Human hallucinations such as:

"I know I put my car keys on the shelf, but they aren't there."
"I recently read a study that said 4 out of 5 Ozempic users experienced migraines."
"No, you said you were going to be there from 7pm on."
etc, etc, etc

The prevalence of human hallucinations is why the watchphrase for historical scholarship (and of Renaissance humanists) is (was) ad fontes - to the sources. It's also why my research group quit releasing preliminary work to friends in the scholarly community with the proviso that everything must be checked before being cited. It turned out that many people couldn't be bothered.