Mike Taylor
@hammer_mt
🧪 Head of Evals @Every
📕 O'Reilly author: Prompt Engineering & DSPy
🎓 400k students on AI Udemy course
🤖 AI engineer since 2020
📈 Built a 50 person agency
3.8K Following 10.2K Followers
Microsoft Copilot and Devin are the only legit enterprise options now if you want to avoid vendor lock-in.
hi can I get 7 more followers so I can be an influencer thanks
AI writing is only allowed if your input tokens > your output tokens.
"Write me an article on X" is slop, "here's an hour of my hard-won experience on X, write it up" is morally defensible and leads to good writing, even if AI wrote the final draft.
Divide input tokens (initial idea plus guidance along the way) by the number of tokens in the final draft to get your distillation ratio.
Show more
Is it just me or is MiniMax H3 Max just not that good compared to the hype for it? P.S. here's a video I made with it of @danshipper surfing the models.
I guess this is as good a way as any to announce that I'm now the Head of Evals at @every
so it looks @google disabled the @youtube account for @every with 0 notice and no reason given...
anyone have any experience with this or know how to help us get it back?
Eliezer on encountering Leopold, which some of you will enjoy:
Some highlights:
• Since DeepMind’s foundation in 2010, the constrained yet rich worlds of games have played a critical role in understanding intelligence
• the spirit of exploration that succeeded in games had profound impacts for other AI systems: AlphaFold applied these foundations to help solve the 50-year grand challenge of protein structure prediction
• For AI research, this is a golden opportunity... demands precisely the capabilities we believe are essential for frontier AI:
• Continual learning
• Memory
• Long horizon planning
• Complex multi-agent dynamics
• Games have always been a mirror for intelligence
@demishassabis @amoufarek @GoogleDeepMind get it.
Show more
babe, what's wrong?
you've hardly touched your claude opus 5
Who do I know that has a Tesla?
- hold farmers criminally liable if they enable meals that endanger children
INBOX: Ken Paxton reveals a data center plan:
- prohibit use of Chinese tech in American data centers
- hold data center companies criminally liable if they enable AI chatbots that endanger children
- cosponsor Sen. Cotton's DATA Act
- support Abbott's state plan re: data centers
Show more
Mordor is in New Jersey, just off the turnpike
Isaac Asimov, explaining how he understood the metaphorical meaning of the One Ring and Mordor in Tolkien's ‘Lord of the Rings’ by driving on a New Jersey highway: ⬇️
Recommend the read!
In ten years I’ll be employed as a contract vintage copywriter by a wealthy billionaire, who flexes by having a human (me) write prose for him
gpt 5.6 sol when you forget to disable internet before running the eval
I was hired to deliver an executive AI workshop and told the CTO wanted to present their own internal tool. Turns out @rnaud had basically built his own Claude Cowork but secured it for healthcare usage. Glad to see this getting profiled, lots of useful learnings for others.
Show more
Everything is a remix
One thing employers need to understand is there is no world in which people stop using AI to do their jobs once they become AI-pilled. The METR productivity study had to stop because they couldn't find enough developers willing to work without AI to run an A/B test.
Show more
Need to start scoring the prompts not the work
Homework scores once predicted exam performance; now those who score highest are, perversely, more likely to do worse in exams.
