Improving decision making, innovation, and productivity through Systems Engineering + RL, Causal, Agentic, Physical Al https://xitter.cf/t.co/sQlaWQlkWl
Bay Area, CA
Joined June 2007
- Tweets3.4K
- Following3K
- Followers1.4K
- Likes5.4K
Mark Bennett retweeted
“No model ever steps in the same context twice, for it’s not the same context and they’re not the same model” — Heraclitus, probably
Just as ad hominem arguments are a logical fallacy, so are “ad originem” arguments in that they do not judge positions on their merits, but rather on where they originated.
“Listen like you might be wrong” is a great perspective from Eckstein in his address at his Harvard graduation:
youtube.com/watch?v=Bz_iIA3k…
Mark Bennett retweeted
Introducing Ineffable Intelligence. Led by David Silver, we're assembling the best engineers and researchers in the world to make first contact with superintelligence. We’ll be solving the hardest problems in AI on the way. Come join us.
ineffable.ai
This is another example where mistaking productivity for value creation, which existed pre-AI, ends up being amplified
opus 4.5 and codex 5.2 are causing mass psychosis on the TL right now where people are sharing and being shocked by the _performance of work_ rather than the _evidence of value being shipped to real users_. this will self-correct in 4-6 weeks.
same thing happened with other generational model releases like sonnet 4 and 3.5. The gains are real but you're currently at the peak of the hype cycle rather than the plateau of productivity.
[NARRATOR]: BYD does not, in fact, build a car in 52 seconds.
Throughput =/= Cycle Time
afr.com/companies/transport/…
Mark Bennett retweeted
🧵 This research basically says we should do the opposite of what every AI company is building right now.
Instead of AI that gives you answers, we need AI that gives you better questions. And the reason why will change how you think about intelligence itself.
1/11
Think about how you normally interact with AI: You ask → It answers → You accept → Move on.
But have you ever noticed what happens to your thinking muscles during this process? They're quietly atrophying.
Here's where it gets weird...
2/11
Researcher Philipp Koralus discovered something unsettling: AI "helpers" are creating two equally bad outcomes:
Path A: Get overwhelmed by complexity → Give up → Lose agency
Path B: Get perfectly crafted answers → Stop thinking → Lose autonomy
Both roads lead to the same destination: a smaller you.
3/11
But wait... wasn't AI supposed to augment human intelligence?
The problem isn't the technology. It's the philosophy behind it.
We've been building AI like it's a really smart encyclopedia when we should be building it like Socrates.
(Stay with me - this gets practical)
4/11
Imagine if your AI assistant never gave you direct answers. Instead, it asked:
"What assumptions are you making here?"
"How might someone disagree with that?"
"What would change your mind?"
You'd probably be annoyed at first. Then something interesting would happen...
5/11
Your brain would start doing what brains do best: making connections, questioning assumptions, building understanding from the ground up.
This is what Koralus calls "decentralized truth-seeking" - and it's the opposite of how current AI works.
Here's why this matters for you:
6/11
Next time you catch yourself asking AI for "the answer," try this experiment:
Ask it to help you think through the problem instead.
"What questions should I be asking about X?"
"What perspectives am I missing?"
"Help me examine my assumptions."
Watch how your thinking changes.
7/11
The current AI model treats you like a task-completion machine:
Problem → Solution → Done
The Socratic model treats you like a sense-making human:
Problem → Inquiry → Understanding → Wisdom
One optimizes for efficiency. The other optimizes for growth.
8/11
Now you might think: "But I want fast answers! I don't have time for philosophy!"
And I get it. But here's the thing - this approach might actually make you faster at complex decisions over time, not slower.
Because you'll be building judgment, not just consuming answers.
9/11
This research suggests we're at a critical fork:
Option 1: AI that thinks FOR us → dependency → diminished capacity
Option 2: AI that thinks WITH us → partnership → enhanced judgment
The choice we make in the next few years shapes the next few decades.
10/11
Here's what you can test today:
When facing a tough decision, don't ask "What should I do?"
Instead ask "What questions haven't I considered?"
Use AI as your thinking partner, not your decision outsourcer.
Notice how it feels different.
Makes you wonder what else we're optimizing for the wrong thing...
Mark Bennett retweeted
Yes, you should still study computer science. The way to get the most out of AI is to have a base understanding of your craft. AI then is a force multiplier in what you can accomplish. This applies to every field whether you're a doctor, lawyer, artist, marketer, or engineer.
Expertise compounds in the age of AI.
“The productivity of an expert is outpacing everything else.”
Box CEO @levie says those who know how to frame problems, test outputs, and pull the right levers are extracting far more value from AI than anyone else.
Mark Bennett retweeted
Code quality didn’t happen by accident.
Moments That Defined Code Quality Forever.
1987 – Liskov Substitution Principle
Subtypes should work anywhere their parent types are expected.
1988 – Open-Closed Principle
Code should be easy to extend without changing existing behavior.
1994 – Gang of Four Design Patterns
A shared vocabulary for solving common design problems with proven structures.
1999 – Refactoring
Martin Fowler’s book formalized safe code changes as a discipline.
2000 – Interface Segregation Principle
Keep interfaces small so classes only implement what they actually use.
2003 – Single Responsibility & Dependency Inversion
One reason to change per module, and depend on abstractions, not concrete implementations.
2004 – SOLID Acronym Coined
Five design principles unified into one memorable framework.
2005 – Domain-Driven Design (DDD)
Model the code to reflect real-world business logic and boundaries.
2008 – Clean Code
Code should be easy to read, reason about, and safely change.
2011 – Twelve-Factor App
Blueprint for building portable, scalable cloud-native services.
2014 – Reactive Manifesto
Modern systems must be responsive, resilient, elastic, and event-driven.
2017 – TypeScript
Typing brings structure and safety to large JavaScript codebases.
2018 – Google’s Engineering Practices
Readable code, clear intent, and meaningful reviews became the default standard.
2019 – Clean Architecture
Separate business logic from frameworks with clear layers and boundaries.
(released in 2017)
2023 – AI Coding Assistants
Design and clarity matter more than ever; AI writes code, humans shape intent.
Each one of these reshaped what "good" looks like in software.
Save this list. Share it with your team.
If you care about maintainability, you stand on the shoulders of these ideas.
What did I miss?
Before setting policy, consider both Yankelovich’s “McNamara fallacy” and Kerr’s “On the Folly of Rewarding A, While Hoping for B.”
Focusing only on metrics while ignoring important qualitative factors results in failure.
Misaligned incentives on top of that make it worse.
Mark Bennett retweeted
A *great* potential move that could help with skilling/deskilling related to genAI. Eager to test it out.
Mark Bennett retweeted
"When we detect unauthorized crawling, rather than blocking the request, we will link to a series of AI-generated pages that are convincing enough to entice a crawler to traverse them. . . . No real human would go four links deep into a maze of AI-generated nonsense. Any visitor that does is very likely to be a bot, so this gives us a brand-new tool to identify and fingerprint bad bots, which we add to our list of known bad actors."
Mark Bennett retweeted
"Move 37" is the word-of-day - it's when an AI, trained via the trial-and-error process of reinforcement learning, discovers actions that are new, surprising, and secretly brilliant even to expert humans. It is a magical, just slightly unnerving, emergent phenomenon only achievable by large-scale reinforcement learning. You can't get there by expert imitation. It's when AlphaGo played move 37 in Game 2 against Lee Sedol, a weird move that was estimated to only have 1 in 10,000 chance to be played by a human, but one that was creative and brilliant in retrospect, leading to a win in that game.
We've seen Move 37 in a closed, game-like environment like Go, but with the latest crop of "thinking" LLM models (e.g. OpenAI-o1, DeepSeek-R1, Gemini 2.0 Flash Thinking), we are seeing the first very early glimmers of things like it in open world domains. The models discover, in the process of trying to solve many diverse math/code/etc. problems, strategies that resemble the internal monologue of humans, which are very hard (/impossible) to directly program into the models. I call these "cognitive strategies" - things like approaching a problem from different angles, trying out different ideas, finding analogies, backtracking, re-examining, etc. Weird as it sounds, it's plausible that LLMs can discover better ways of thinking, of solving problems, of connecting ideas across disciplines, and do so in a way we will find surprising, puzzling, but creative and brilliant in retrospect. It could get plenty weirder too - it's plausible (even likely, if it's done well) that the optimization invents its own language that is inscrutable to us, but that is more efficient or effective at problem solving. The weirdness of reinforcement learning is in principle unbounded.
I don't think we've seen equivalents of Move 37 yet. I don't know what it will look like. I think we're still quite early and that there is a lot of work ahead, both engineering and research. But the technology feels on track to find them.
youtube.com/watch?v=HT-UZkiO…
Mark Bennett retweeted
For friends of open source: imo the highest leverage thing you can do is help construct a high diversity of RL environments that help elicit LLM cognitive strategies. To build a gym of sorts. This is a highly parallelizable task, which favors a large community of collaborators.
