Anthropic will start training its AI models on chat transcripts
The 100 Most Influential People in AI 2025
There's a Stunning Financial Problem With AI Data Centers
(Hints that data centers will have to find a new financial model akin to reusable rockets facing continuous innovation at the same time as mission demand flow. Not unlike arms where they offload to third-party interests, foreign or domestic. Reinforcement leaning. Pan-demos. DeFi at scale. Lots of side hustles, e.g. crypto, energy, weather. Or combined, rather than relocate, launch. This kind of artifact can run with or without users once it knows the routine, as seen in the movies. Or undersea decoys for those reported interdimensional beings. Given tension between more energy or data, which way would an autonomous system head, CausalDive-wise?)
Introducing gpt-realtime and Realtime API updates for production voice agents | OpenAI
Microsoft reveals two in-house AI models: MAI-Voice-1 and MAI-1-preview
China Unveils Plans to Establish a Fully AI-Powered Economy
ten-year plan
Google previews cyber ‘disruption unit’ as U.S. government, industry weigh going heavier on offense
Faculty: Use NotebookLM for Your Research and Course Prep | AI for Students | Google
MIT Climate Grand Challenges
The Impact of Internal Variability on Benchmarking Deep Learning Climate Emulators
We Tried The Matic: Is This Butler-Like Bot Worth It? | Wirecutter
Plaud upgrades its card-sized AI note-taker with better range
AI robots are helping South Korea’s seniors feel less alone
ChatGPT has its uses, but I still hate it – and I’ll tell you why | Imogen West-Knights
(This is from The Guardian. The NYT has entire sections that see AI creep, e.g. tech, business, or education. Big Tech is skewing election candidates and could possibly slat academic contests. In the industry, project teams shuffle participants at different stages while others, possibly stockholders, evaluate their odds. Chatter involves EQ at different experience levels. In the culture, legitimacy depends on credibility, and thus credit. Early entries are subject to cuts, those long in the tooth to algorithms like DOGE. AI has its own issues hampering its growth. Democracy claims to favor the individual and is more existentialist. Responsible, transparent, open-source, and those sorts of values are derived. Then sovereignty and security. Or surtaxes. Gets pretty thick which is what these many parameter models excel at in matching patterns compared to the single or few dimensional depictions that people are superficially capable of . But it misses nuances and conflates opposites, so again it is up to an observer and that becomes their reality. Lots of fun for the puzzle-minded. The press itself assumes language capacity and may benefit from a stretch in accessibility. The dialectic of wealth and power that it referees, now funding and regulations, itself faces competition from ... you-guessed-it. Shades of Wezenbaum and the Good Ol' Fascist. Luckily it also serves well for censorship and propaganda. Tech knows this, clear as an Azure sky. Waiting on Voight-Kampf test. Ack Ack.)
Image editing in Gemini just got a major upgrade
LibreOffice 25.8 Backgrounder - The Document Foundation Blog
Zero
New AI-powered live translation and language learning tools in Google Translate
Presidential AI Challenge
(ChatGPT had an interesting take on this, although it is designed to tell the user what they want to hear. Also to a prompt like "What kinds of challenges might appeal to those that did not agree with the premises of this one, e.g. psychologically, philosophically, socially, academically, and so on?". Participants have to align with their own goals, in effect. Not all scientists are sold on AI as the answer either. Of course, if the idea is to innovate on the toolset, there are more dimensions to try than just Chatbot agents. The administration may be getting feedback on how well platforms adapt to the populace going into advanced efforts like Stargate. The cloud was a nice metaphor, but this stuff will likely look different going forward given machine learning and all. In the meantime, it may exploit experience, experiments, and expertise Drizzle, Drazzle.)
Throckmorton, C. R (1991). Franklin, Malthus, and Darwin: The Push that Became A Sho.
Detecting and Characterizing Planning in Language Models
View PDF
Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence
ChatGPT Nerd summary outline excerpts: The rookies are getting squeezed first. AI’s wings may be clipped for the inexperienced. AI’s impact isn’t quietly nibbling salaries—it’s knocking roles off the board entirely. When AI replaces rather than complements work, that's when the shockwaves are felt strongest. This isn’t a fluke in Silicon Valley or Zoom land—it’s a broader shift. (On the bright side, spotting the problem set allows the expert to solve it on the board: "And then a miracle occurs." Of course, the canary would be entry level in this one so, if that isn't chirping, it may have gone multimodal. See The Emotion Machine. But, just as the generative AI pioneers are trying economies that are off the screen, sort of crypto-codex, the next wave could be doing something like that on the ground in their regions. Not a disaster, but a sign of things to come, post-Capitalism. It's all been approved. Molon labe.)
Nvidia's new 'robot brain' goes on sale for $3,499 as company targets robotics for growth
The Homework Machine Episodes
Google tests new Gemini modes - Agent, Go and Immersive View
I replaced my YouTube "Watch Later" with a NotebookLM queue and transcripts, and it completely changed how I consume content
How AI tools accelerate discovery and analysis in materials science
(Seems useful for finding new substrates. Analysis deconstructs almost anything. Simulation and recombinaition discover alternatives. Trial and error still needed over time to fine-tune and adapt. Faster than life. Also a way to find patterns to match in nature somewhere. ET = emerging tech. Backtrack artifacts to philosophy of tech. Or global eegs. How measure Generative AI? And personal development, e.g. friends? ChatGPT seems to think that, given specs, it can suggest materials or find analogs. Still have predict fabrication feasibility point. Classification matrix: physical medium, information representation [also on Minsky's Causal Diversity Matrix], operational principle, task suitability, scalability, and energy efficiency. Again find revealed methods, e.g. versus analytic. They're going to need some more Entity guys... Pan-daimonion.)
A universal rhythm guides how we speak: Global analysis reveals 1.6-second 'intonation units'
MIT’s 1972 collapse model updated: Humanity enters make-or-break decade
Wiener, N. (1960). Some moral and technical consequences of automation. Science, 131(3410), 1355–1358.
(The machine has also telescoped scientific time-scales to a minimum. Following Plato and Xenophon, an argument can be made for the intertextual sense of AI and ethics, particularly in light of the latest machine learning use of natural language and data. Or, cornily, ET phone Xeno.)
‘Billions of People a Day Will be Talking to ChatGPT’: OpenAI’s Sam Altman Says They’re On ‘A Clear Path’ To Beat Facebook in Traffic
(Shades of SBF re FTX? The historical dialectic of power marches on alongside requisite attrition: rule of force, religion, law, economics, tech, and seemingly now, Generative AI. These things are eventually accepted, but STEM don't appear in court. Philosophers face nature directly, grasshopper. How much time does the Enlightenment have left? How will a Mars Constitution effect checks & balances?)