AI Opinions (6)

AI News (8)

AI Research (2)

A former employee writes about the danger of Palantir
Whistleblowing - This will probably make the rounds today, if for nothing more than the “AI kill-chain” phrasing used. In short, a former Palantir technical writer calls out the company’s platforms which combine personal and public data to track sources. They imply that these systems might be violating human rights / constitutional rights. Not a bad read, but there are better write-ups on surveillance state concerns. Not much there there in this one.

!!! AI leaders talk collaboration-thinking over automation-thinking
AI Advocacy - MIT’s David Autor and Google’s James Manyika argue AI should be built (designed from the start) to collaborate with professionals, not replace them. Quote of the day: “Imperfect automation is not a first step toward perfect automation, anymore than jumping halfway across a canyon is a first step toward jumping the full distance. Recognizing that the rim is out of reach, we may find better alternatives to leaping”

Is the AI bubble about to burst and collapse the market?
AI Concern - With AI stocks sliding at the tail of last week (we talked about this), Guardian columnist Phillip Inman compares the frenzy to the dot-com crash. He ends with the caution that AI will probably be bad for humanity, but great for investors.

AI is changing work, not stealing it from us
AI Advocacy - AEI fellow Bronwyn Howell counters the fears of mass unemployment resulting from AI. It’s a quick read and draws on some nice, tangible examples of past technological innovations that didn’t exactly leave us longing for more to do.

Agentic AI and our willing handover of personal data
AI Advocacy - Another from AEI, this time a podcast, Shane Tews and Imbue’s Matt Boulos unpack “agentic AI”. They warn that without interoperability, we could end up with locked-down “super-apps” trading privacy for convenience.

Are we heading for another AI winter?
AI Advocacy - Harvard’s Paulo Carvão revisits past AI winters to ask if history might repeat. The general takeaway is that, no, the AI winter is not coming again, mostly because this version of AI is already so integrated into systems and culture. Too much momentum to stop.

!!! IBM & NASA release Surya for solar-flare prediction
IBM and NASA launched Surya, an AI model that predicts solar storms with better accuracy and speed by modeling the Sun as a digital twin. I learned the word “heliophysics” from this writeup, which is also cool. Pretty much every modern comfort somehow relies on solar flairs not randomly knocking us out, so predicting them ahead of time = good. Now open-source.

Apple in talks for Gemini-powered Siri
Apple is exploring licensing Gemini from Google for a Siri upgrade. They also looked at Anthropic’s Claude and OpenAI models earlier this year.

Meta partners with Midjourney for AI aesthetics
I pay for a Midjourney account and use it for work, bedtime stories, side projects, this newsletter…. all sorts of stuff. Sounds like Meta secured a deal to license Midjourney’s “aesthetic” tools for its product line. Expect a lot more variety and less AI-slop looking images in Instagram and Facebook. I’d expect this to somehow find its way into Marketplace as well.

Waymo gets the go-ahead in NYC
As an Austin resident, I’m used to a good Waymo annoying me with its safe turns in an intersection. Now, Waymo is the first company approved to test autonomous vehicles in New York City. Striking, considering existing state laws regarding human-drivers and the obvious Taxi industry hurdles they must have faced. If the rollout goes well, it could reset how cities approach AV regulation and deployment.

Sferical AI launched by Swedish Conglomerate
The Swedish Wallenberg foundations unveiled Sferical AI, aimed at building Europe’s own industrial-scale AI infrastructure. Mainly including this because I’m repeatedly impressed by Swedish tech lately (Spotify, Lovable, Leap.ai) and expect this to create even further gravity there.

Musk tried recruiting Zuckerberg for OpenAI deal
Court filings show Elon Musk offered Mark Zuckerberg involvement in a $97 billion takeover of OpenAI. Never went through. My only takeaway on this strategy is: “okay so it’s like when Daenerys or Jon temporarily join forces to take out the more dominant Cersei/Lannister seat — knowing they’ll have to later fight again to settle who the real ruler is”

South Korea looks to AI to fix economy, big investment in culture AI
Seoul pledged tax incentives and a 100 trillion-won fund for AI and innovation, launching 30 government-backed projects. Particularly interesting to me is the emphasis on using government-led AI investments to specifically drive the cultural sector (music, film, ‘web toons’) and create more international interest in those spaces, through integrated AI.

OpenAI establishes first India office
While pricing hurdles are a notable concern, OpenAI is looking to build a big footprint in India. From the article: “It plans to get feedback from Indian users to make its products relevant for the local audience and even build features and tools specifically for the country.”

OpenAI designs Yamanaka proteins with GPT-4b
Partnering with Retro Biosciences, OpenAI used a model to design protein variants that boosted cell reprogramming 50 times better. A “Yamanaka” protein is a protein that can turn regular cells into stem cells for all sorts of powerful applications. Think, blindness, organ failure, diabetes, aging… This use of AI helped design better proteins that are far more effective in lab settings. From the article: “When researchers bring deep domain insight to our language-model tooling, problems that once took years can shift in days”

AI scribes ease doctor burnout
A JAMA study found AI “scribes” (note takers for doctors, like in surgical settings) halved burnout rates at Mass General Brigham and significantly boosted positive feelings (doctors’ sense of wellness) at Emory. The cool thing here is that the research points to measured benefits to the user, beyond the task automation, when applying AI to partner in work.