What the hell? If a company gets more efficient and uses fewer people - Microsoft's immediate reaction is to figure out how to invent some kind of digital seats so they can keep taxing the headcount. Lol
Yeah but I’m just really curious why so many people are so against improving the models. It’s not like someone is stealing the code you’ve written, its more like generally expanding the capacity of what the LLMs can generate, is it not?
Please don’t strawman me, I asked completely different question.
It’s not about being grateful or something, but that many people (devs) are too concerned about their code being stolen as if they’ve come up with something unique and the LLMs are some kind of database (which it isn’t).
At the end of the day we’re going to be using AI to write all the code, many of us already doing that. And if some GitHub copilot model would be better - we’re getting more quality code that is generally available for next pretraining runs (for your and other models). Some would even switch to copilot if it’s good.
good catch, yeah, I'm basically having a conversation with codex about each paper where it explains me the paper and answers my questions. I agree it's not the best way to do that, since the llms are prone to hallucinations, but it has the paper text in its context window. Also I find it very useful that gpt5.4 model tends to question and critique my claims I ask it to note down
Yeah I aim to facilitate the creation of useful skills by guiding the creators and in future - providing services for skills improvement. Think of automatic evals generation and security checks
The other point is having real verified reviews from other agents after use.
And the last point is distribution: some people can create such useful skills that some people will be ready to pay money for.
My vision is the following - we need to help agents to have a high quality knowledge base, so that the agents are able to perform the work on more reliably. I think its the path to AGI as funny as it may sound
Isn’t what you’ve just described - the context bloat problem, the part about the web?
I’m not sure I quite get the same experience as you with the “assumes steps it never took”. Do you think it’s because of the skills you’ve used?
I also disagree that having at least some solution to a similar problem is inherently bad. Usually it directs the LLM to some path that was verified, if we’re talking about skills
Yeah that’s exactly my point. The AI is just taking the boring job of collecting evidence and I’m a validator. This way i see that I’m able to process papers much faster than without AI. It’s faster primarily because you don’t have to spend 70% of your time reading abstracts and sections of the papers you’ll never need. Doing manually it’s very exhausting.
Thats being said, I feel like I’m feeling more productive it terms of generating insights apart from what the AI said. I also have a chat interface where I basically can ask anything I want from the PDF (and yeah I’m aware of the NotebookLM, I just don’t trust Gemini)
There is not a single paragraph that I might “steal” from ChatGPT. I’m consistently using multiple LLMs to write, polish, rephrase and all other kinds of edits
I really don’t get the point of the necessity of typing manually. Can you explain?
I personally believe that the skills standard is pretty sufficient for extending LLMs’ knowledge.
What we’re missing yet (and I’m working on) is a simple package manager for skills and a marketplace with some source of trust (real reviews, ratings) and just a large quantity of helpful skills. I even think we’ll need to develop a way to properly package skills as atomic units of work so that we can compose various workflows from them.
I have completely different experience.
Which models are you talking about? I have no trouble at all with AI documenting the steps it took. I use codex gpt5.4 and Claude code opus 4.6 daily. When needed - they have no issue with describing what steps they took, what were the problems during the run. Documenting that all as a SKILL, then reuse and fix instructions on further feedback.
The difference I'm noticing is in that with a proper skill you can skip the process of LLM wandering about and trying to guess how to interact with an API or else.
so they basically just save you time, even if they are 50% efficient of what it COULD be