They're giant pattern regurgitators, impressive for sure, but they only can be as good as their training data, reason why they seems to be more effective for TypeScript, Python etc. Nothing less nothing more. No AGI, no Job X is done. Hallucinations are a feature, otherwise they would just spit out training data. The thing is the whole discussion around these tools is so miserable that I'm pondering the idea of canceling from every corner of the internet, the fatigue is real and pushing back the hype feels so exausting, worse than crypto, nft and web3. I'm a user of these tools me pushing back the hype is because its ripple effects arrive inside my day job and I'm exausted of people handing to you generated shit just to try making a point and saying "see? like that"
I don't understand how Agents make you feel productive. Single/Multiple agents reading specs, specs often produced with agents itself and iterated over time with human in the loop, a lot of reviewing of giant gibberish specs. Never had a clear spec in my life. Then all the dancing for this apperantly new paradigm, of not reviewing code but verifying behaviour, and so many other things. All of this to me is a total UNproductive mess. I use Cursor autocomplete from day one till to this day, I was super productive before LLMs, I'm more productive now, I'm capable, I have experience, product is hard to maintain but customers are happy, management is happy. So I can't really relate anymore to many of the programmers out there, that's sad, I can count on my hands devs that I can talk to that have hard skills and know-how to share instead of astroturfing about AI Agents
To me part of our job has always been about translating garbage/missing specs in something actionnable.
Working with agents don't change this and that's why until PM/business people are able to come up with actual specs, they'll still need their translators.
Furthermore, it's not because the global spec is garbage that you, as a dev, won't come up with clear specs to solve technical issues related to the overall feature asked by stakeholders.
One funny thing I see though, is in the AI presentations done to non-technical people, the advice: "be as thorough as possible when describing what you except the agent to solve!".
And I'm like: "yeah, that's what devs have been asking for since forever...".
With "Never had a clear spec in my life" what I mean is also that I don't how something should come out till I'm actually doing it. Writing code for me lead to discovery, I don't know what to produce till I see it in the wrapping context, like what a function should accept, for example a ref or a copy. Only at that point I have the proper intuition to make a decision that has to be supported long term. I don't want cheap code now I want a solit feature working tomorrow and not touching it for a long a time hopefully
In my real life bubble, AI isn't a big deal either, at least for programmers. They tend to be very sceptical about it for many reasons, perceived productivity being only one of them. So, I guess it's much less of a thing than you would expect from media coverage and certain internet communities.
Just because you haven't or you work in a particular way, doesn't mean everyone does things the same way.
Likewise, on your last point, just because someone is using AI in their work, doesn't mean they don't have hard skills and know-how. Author of this article Mitchell is a great example of that - someone who proved to be able to produce great software and, when talking about individuals who made a dent in the industry, definitely had/has an impactful career.
Hell I see the big banner picture hallucinated by a prompt and all I see is an unproductive mess. Won't comment on the takes the article makes they're just miserable
There is little to no integration between deterministic IDE features(like refactorings) and LLMs. For example I don't want a statistical tool to rename a method by predicting tokens, I want it to use IDE features and not via another higher abstraction protocol like mpc, I want deeper integration. Sometimes I look at comments in code and think "why can't I have an agent checking if the content of a comment actually reflect the code below?" I feel like we're light years away from a killer integration
This might actually be another area language servers shine. As I understand it, the TS Language Server can do renames. Ergo, we ought to be able to have the LLM ask the lang server to do the rename instead of trying to do it itself. That'd be easier than trying to integrate with each IDE individually. (Whereby "IDE" seems to be synonymous with "VSCode" lately...)
Agree, another improvement i'd like along the lines or renames is lsp suggestions for method names, enums, functions, etc The llm should be able to autocomplete given lsp available symbols, this way it would avoid far less hallucinated methods
After using CC in VSCode a bit, I find it makes liberal use of Pylance, both proactively (in-thread) as well as during the lifecycle (guessing via hooks). It’s almost annoying how, and I’d like to figure out how to get it to use repos lint rules.
Deep integrations are hard and the AI companies are just winging it when it comes to eating their own dog food. Their apps are bare bones, somewhat flaky, and overall not that impressive from a UX point of view.
It's very obvious that while their AI teams are top notch, their product teams are very middle of the road. Including design. Even though they apparently engaged Jony Ive, I can't actually see his 'touch' on anything they have. You'd expect them to have a much higher level of ambition when it comes to their own products. But they seem stuck getting even the basics shipping. I use Chat GPT for Desktop. It's alright but it seems to have stagnated a bit and it has some annoying bugs and flakiness. Random shit seems to break regularly with releases as well.
Another good example of the lack of vision/product management is the #1 and oldest use case for LLMs since day 1: generating text. You'd expect somebody to maybe have come up with the genius idea of "eh, hmm, you know, I wonder if we can do better than pasting blobs of markdown rendered HTML to and/from a word processor from a f**ing sidebar".
Where's the ultimate agentic word processor? The ultimate writing experience? It's not there. Chat GPT is hopelessly clumsy doing even the most basic things in word processors. It can't restructure your document. It can't insert/delete bits of text. It can't use any of the formatting and styling controls. It can't do that in the UI. It can't do that at the file level. It's just not very good at doing anything more than generating bits of text with very basic markdown styling that you might copy paste to your word processor. It won't match the styling you have. Last time I checked Gemini in Google docs it was equally useless. I don't have MS Office but I haven't heard anything that suggests it is better.
For whatever reason, this has not been a priority (bad product management?) or they simply don't have the creativity to see the rather obvious integration issues in front of them.
Yes making those is a lot of work and requires a bit of planning. But wasn't the point of agentic coding that that's now easy? Apparently not.
I don't know you, but apart from
ai tools race fatigue(feel pretty much like frameworks fatigue), all I see is mouse traveling a lot between far distant small elements, buttons and textareas. AI should have brought innovation even in UIs we basically stopped innovating there
Did we really need that? I wonder if people recognize how bad the "standard committee" is and that we are held back by them. I can't hold in my mind how many features the web is missing that should be prioritized, I think I'll make a list and start a blog. The web deserve better
> Unsurprisingly, participants in the No AI group encountered more errors. These included errors in syntax and in Trio concepts, the latter of which mapped directly to topics tested on the evaluation
I'm wondering if we could have the best of IDE/Editor features like LSP and LLMs working together. With an LSP syntax errors are a solved problem, if the language is statically typed I often find myself just checking out type signatures of library methods, simpler to me than asking an LLM. But I would love to have LLMs fixing your syntax and with types available or not, giving suggestions on how to best use the libraries given current context.
Cursor tab does that to some extent but it's not fool proof and it still feels too "statistical".
I'd love to have something deeply integrated with LSPs and IDE features, for example VSCode alone has the ability of suggesting imports, Cursor tries to complete them statistically but it often suggest the wrong import path. I'd like to have the twos working together.
Another example is renaming identifiers with F2, it is reliable and predictable, can't say the same when asking an agent doing that. On the other hand if the pattern isn't predictable, e.g. a migration where a 1 to 1 rename isn't enough, but needs to find a pattern, LLMs are just great. So I'd love to have an F2 feature augmented with LLMs capabilities
Don't think so, and we should stop spread damaging narrative like this. I'd say it's already able to imitate this kind of explainers(badly) thanks to his training data. All the subtle teaching nuances, effort, know-how and visual creativity that people like Bartosz Ciechanowski put on this kind of work is not reproducible if not statistically imitating it
reply