It's not an appropriate solution to for AI vibe coding IMO. It's too complex. People without a technical background need a simpler abstraction than RLS.
Coming soon: language models trained on 2000 years of tactile, proprioceptive, spatio-temporal, calorific, olfactory, gustational, humidity, temperature and barometric data.
The downside: the model doesn't always have words to describe what it's feeling.
Ask the model to declare its assumed axioms. Agree that those can be trusted for the time being. Things like the behaviour of third party libraries. Ensure that your review agent enforces this.
Thanks for the advice. This is what I struggle with the most. I’ve learned formal verification other than Lean proofs and used it a few times. It holds up well, but as soon as it becomes a slightly longer-term task, it passes formal verification but turns into work that doesn’t reach the goal G I’m aiming for, and I don’t know how to solve that. I clearly need a separate review agent—have you built a separate harness for the review agent?
Other people say that when they do AI coding, they finish quickly, but for me it’s painful because I keep having to verify it myself.
- Learning in a self-directed way that aligns with the needs of your role and where it is going
- Clear communication, the ability to edit and know how to say what is important
- Forming and testing hypotheses
- Evaluating and testing tools and work product, and being able to recognize and rectify deviations
- Delegating, setting constraints, rubrics and escalation criteria
- Having a clear definition of done and being able to stick with it
- Understanding business domains and modelling them usefully
- Collaborating with others, communicating your needs and capabilities clearly, eliciting the needs and capabilities of others, synthesizing conflicting needs and aligning compounding capabilities
The adoption of genAI and agentic tooling really just serves to make all of these more important.
I basically agree with you, even though I think AI is a bad thing. But I think your list should be qualified. AI itself modifies expectations and changes the nature of the game, so that people care less about rigor in the way you described. With AI, since it's so easy to make new apps, people will care less about whether an app is correct or well-defined. And over time our expectations will change and our reactions will be gamed so that the person who makes the app the quickest with the most flash will win over those who put more care into things.
There's always been a tension between rigour and marketing of course but AI is a new order of magnitude shift towards the marketing and hype side of things. So I think while your skills are still needed, they will be overshadowed by a new set of skills to manipulate the audience or userbase to use the suboptimal more than was possible before.
I guess I am not doing my PhD in an actual research lab then. We don't get Claude/chatgpt paid, but we basically have to use them.
Otherwise we don't have any chance for sota research. And I am not talking about AI doing innovation. Just shorten the time between idea and results.
Right now we have to Pay it ourself, and we are very aware that oai might use our research.
It has already been shown that this is a fig leaf. Users that did not consent to sharing their data have had their images uploaded to external services without their consent.
The users in this case had given permission for their data to be used for training (or at least hadn't opted out). They had not, of course, given permission for their data to be shared publicly or with a third party in this way. Consequently, OpenAI categorized this as a breach (i.e., something they did not intend to do), and sought to rectify the situation.
Why would you assume that when they are violoting the very social contract that underpin the free software movement? I have published my oss work under the AGPL software that's used in production by millions of people because of Stallman's stance on free software, the very deal that is if you benefit from my code and use it from something, you must be open sourcing your code too and can't keep it behind close doors
> I think it unlikely given the harm it would do to them
Did it harm them in any real way that the original training material was also of questionable origin? Not the least.
We're being naive if we think this is not happening. Apart from some high profile enterprise contracts (which average Joe Physics Researcher does not have as they are using "free" credits insteador pay a personal account out of their own pocket), everything remotely attractive is being mined. Too sweet of a gold mine to pass by.
I'll stand corrected if an official statement from OpenAI/Anthropic comes out stating the opposite together with a whistleblower program worth its name. Until then it is safe (and reasonable) to assume that this is going on. The silence is telling.
It could be that some researchers can get access to latest models or no limit accounts only if they approve of usage of their data. It would be interesting to know which researchers have had which settings/setups.
How someone can prove it when they are lying? How someone can audit the fact that zero data retention enforced or even more that zero learning from this process occur? not only about discoveries them selves but also about the dropped investigation paths without successful leads?
> but I think it unlikely given the harm it would do to them to violate an enterprise contract in this way
That's if you have any reason to believe in the justice system to actually enforce that, which has already been proven complicated (although not impossible!) to demonstrate for copyrighted content.
At this point believing the frontier labs at their word is utterly naive. They have cheated and they will cheat if it gives them any hope of leaping ahead of their competitors.
OpenAI (as others) has a Researcher Access Program [0] which grants ChatGPT credits for research institutions and individuals.
They review applications beforehand to assess the "relevance of the research proposal" and check if it fits "the objectives and research priorities for this program".
There is no statement whatsoever that the data provided will not be used for training (so there would not be an "lying" required), but I'm happy to be proven wrong.
In essence, they have a process in place to grade and classify the type of information coming from each of those accounts prior to their enablement. I see no issue for them to feed that data back into training, even explicitly weighted for the field the research-access was given to.
Honestly, it's a perfect process: "Physics-research grants are reaching critical mass. Let's put priority on more physics field grants and boost the compute-resources for the physics model. Let's aim for some press-releases end of next quarter."
youtuber vibe codes a transpiler says they're not going to bother maintaining it - but big company like cloudflare takes it seriously enough to absorbe into their own official services.
What is happening?
What is even software dev anymore?
What didnt the labs or cloudflare do this themselves, I mean they did, but then if it's all vibe coded, why do they need someone else's vibe coded stuff?
> why do they need someone else's vibe coded stuff?
Because even if vibecoded, the initial work has already been done, and it would be absolutely foolish to outright dismiss it just because it doesn't meet puritanical standards.
> How is any of this serious?
It isn't & shouldn't be. The upfront demand for "serious" is what kills public projects in the first place, where they all come from an experimentation perspective.
There is no public project out there that (from its first commit) was fully serious: Each & every one of them were offshoots & experiments, testing to see if the thing was possible in the first place.
reply