clbrmbr

10 ms·

181.
by clbrmbr 1y ago
Nicely done. I particularly like the emphasis on writing specs which really is something new in the space and makes Kirk not just “Cursor clone”. This is something missing in Claude Code… the user needs to remember to ask Claude to update t
182.
by clbrmbr 1y ago
Do you think constitutional approaches would help here? (Verifiable reward for the main score, but then asking the model to self-critique for security and quality.)
183.
by clbrmbr 1y ago
Can I start with a natural language description of the theorem to be proven and the model will automatically formalize it? And can I get a natural language interpretation of the Lean proof in case of success? --- I'm thinking of workin
184.
by clbrmbr 1y ago
Does anyone have experience getting agents to understand terminal applications? Like, in general an arbitrary ncurses application. A more specific case I’ve struggled with is output from a long-running program like ping. You’ve got to know
185.
by clbrmbr 1y ago
I hope you are right. We need really impactful failures to raise the alarm and likely a taboo, and yet not so large as to be existential like the Yudkowsky killer mosquito drones.
186.
by clbrmbr 1y ago
I agree that the strongest agentic models (Claude Opus 4 in particular) change the calculus. They still need good context, but damn are they good at reaching for the right tool.
187.
by clbrmbr 1y ago
Suggestion: add a -p option: spegel -p "extract only the product reviews" > REVIEWS.md
188.
by clbrmbr 1y ago
The definition there is standard undergraduate computer science theory. Maybe not standard for software engineering though.
189.
by clbrmbr 1y ago
Nice! I’ve wanted this for years. Suggestion: you may be able to integrate SRS into the conversation. —- you could encourage the model to use certain words, and more importantly you can track the student’s active use of words that are on th
190.
by clbrmbr 1y ago
Fast following is a reasonable strategy. Anthropic provided the existence proof. It’s an immensely useful form factor for AI.
191.
by clbrmbr 1y ago
I would think that rather than ifdefs, one could use separate port files. And regarding C89/99, a solution here is to use ANSI C, which is what Lua does.
192.
by clbrmbr 1y ago
Indeed. Anyone who has built things with Claude Code (Opus 4) and/or something more than one-shot with o3 should be feeling the AGI at this point. Certainly there’s still many limitations, but progress is undoubtedly moving forward.
193.
by clbrmbr 1y ago
The finite grid of resistors (or arbitrary impedances) is actually of great practical usefulness.
194.
by clbrmbr 1y ago
But those fall infinitely short of doom.
195.
by clbrmbr 1y ago
Marks’ paper with Max Tegmark “Geometry of Truth” is a great read, and I can see the ideas repeated here. I’ve been meaning to repro some of the geotruth paper….
196.
by clbrmbr 1y ago
Yeah Claude Code (Opus 4) is really marvelous at code review. I just give it the PR link and it does the rest (via gh cli). — it gives a md file that usually has some gems in it. Definitely improving the quantity of “my” feedback, and certa
197.
by clbrmbr 1y ago
You may be above average intelligence. Those challenges are like classic IQ tests and I bet have a significant distribution among humans.
198.
by clbrmbr 1y ago
We can reduce p(doom | superintelligence) though safety efforts.
199.
by clbrmbr 1y ago
On the former a recent post here “LLMs are cheap” I think, laid out a pretty compelling argument that API inference costs are in the right ballpark. Regarding increasing maintenance, i’m finding that Claude code is just as good for working
200.
by clbrmbr 1y ago
Fascinating reading the section about why the 1980s AI industry stumbled. The Moore’s law reasoning is that the early AI machines used custom processors which were commoditized. This time around we really are using general purpose compute t
201.
by clbrmbr 1y ago
Title should end (1999), as 1977 is the birth year of the author not the publication date.
202.
by clbrmbr 1y ago
Unfortunately o3 is blocked by Google scholar from checking the refs. Would be a big plus for a Google agent if it had fast unfettered access to Scholar!
203.
by clbrmbr 1y ago
They really nailed a casual style that didn’t take way from the depth. More publications using the word “brr” please.
204.
by clbrmbr 1y ago
Naive question: why doesn’t the market regulate electricity consumption? I’ve heard the answers pre-AI, but I wonder how this new general purpose use of electricity changes the calculus?
205.
by clbrmbr 1y ago
A few favorites: wine - beer = grape juice beer - wine = bowling astrology - astronomy + mathematics = arithmancy
206.
by clbrmbr 1y ago
Certainly a shame if true, there are some really sharp folks at Anthropic and this is an important building block in the emerging ecosystem.
207.
by clbrmbr 1y ago
Awesome! Reminds me of the good old days of QuickBasic and SCREEN 13, when you could write very small programs with fullscreen graphics. I still have not figured out how to do fullscreen graphics on my Mac.
208.
by clbrmbr 1y ago
Whoah, how odd. It asked me what I was doing, I said I just ate a burger. It then got really upset about how hungry it is but is unable to eat and was unable to focus on other tasks because it was “too hungry”. Wtf weirdest LLM interaction
209.
by clbrmbr 1y ago
100% this. I’d take it a step further and say that sales tax should be included when you are logged in and it can be anticipated, like is the case in most other countries.
210.
by clbrmbr 1y ago
IMO you want a business cofounder who has a tech background, but they have to be a sharp dollars person. Somebody has got to close deals, negotiate hard with suppliers, and keep the lights on. All things that engineers often don’t do so wel

More ›