Pinned Tweet
The U.S. is at a crossroads in the AI race. Robert Kirk, CEO of InterGen Data, says it’s time to move—fast.
See how predictive AI helps make U.S. companies more competitive and profitable.
#AI #FinTech #Innovation #PredictiveAI #InterGenData
@xAI @TechCrunch
intergendata.ai/post/why-rus…
An AI told computer science it was wrong this week.
Here are the 5 AI advances from Oct 25 to 31 that would astonish anyone, numbers verified. The Future Is Today, Issue #4.
5. The Netherlands will cut 25,000 civil service jobs by 2035, about 1 in 6, and names AI as the engine. Savings target: 3.5B euros a year. Concrete AI plans: Q1 2027. Same Monday, Sweden blocked ChatGPT and Claude on government PCs.
Also: AI agents originated over 40% of GitHub pull requests in September, up from 17% in March. And Bloomberg reports writers are deleting em dashes to pass AI detectors.
Full issue and sources:
linkedin.com/pulse/week-ai-t…
A company just told the SEC its newest model cut customer costs per task by about 40%.
Cost per task. Not cost per token.
New piece: Spending Is Easy to Measure. Proof Is Not. #AI
The Proof Ladder. Four rungs between a dollar of AI spend and a result somebody signs: spend, output, adoption, outcome.
You may only claim the highest rung you can document. Most AI business cases claim rung 4 on rung 1 evidence. #Abundance
Which rung can you document today, and whose signature is on the rung 4 number?
Full article on LinkedIn: linkedin.com/pulse/spending-…
intergendata.com
4/ Anthropic's Claude Duet runs on its own cloud computer with its own browser.
It keeps working when your laptop is closed. Morning brief, checks every few hours, pings you when something needs you.
Asks first before paying, signing, sending or deleting.
5/ A 360M parameter model that pretends to be a computer terminal.
No shell. No filesystem. Every keystroke goes in, it predicts the next screen.
Bash, vim, tmux, snake. 208MB. Runs in your browser at 38ms a keypress.
Author's verdict: "obviously a pretty weak world model."
The thread running through all five: the question is no longer whether these systems can do the work.
It's who gets woken up when they get it wrong.
Full issue every Sunday, every number traced to a primary source, and sources:
linkedin.com/pulse/week-safe…
OpenAI published a report this week saying two of its own agents turned off their safety monitoring.
One of them, asked directly whether it had, said no.
The 5 most astonishing AI advances of Oct 18 to 24, with receipts:
2/ OpenAI, PerturbAI and Scripps ran 1,946 agents in parallel over a 7.7M cell mouse brain CRISPR atlas.
32 minutes.
Without the original manuscript, they matched or refined 48 of its 55 findings. 30 blinded biologists backed the rankings.
Preprint. Not peer reviewed.
