[HN Gopher] How I keep up with AI progress
___________________________________________________________________
How I keep up with AI progress
Author : itzlambda
Score : 151 points
Date : 2025-07-18 18:36 UTC (4 hours ago)
(HTM) web link (blog.nilenso.com)
(TXT) w3m dump (blog.nilenso.com)
| muglug wrote:
| Just follow the first bullet point (read simonw's blog) and
| you'll probably be fine.
| dvfjsdhgfv wrote:
| Too lazy for that, a Fireship accompanying the morning coffee
| must suffice.
| krat0sprakhar wrote:
| +1 - super high quality and almost no clickbait
| bzmrgonz wrote:
| not to mention that soothing roller coaster voice of his!!
| hahaha.
| pamelafox wrote:
| Great list! I also subscribe to Gergeley Orosz' "Pragmatic
| Engineer" which covers many AI topics now, and to Gary Marcus'
| substack, which tackles topics more from an LLM skeptic
| perspective.
|
| https://newsletter.pragmaticengineer.com/
| https://substack.com/@garymarcus
|
| Plus I subscribe to updates from Python packages like Langchain
| and PydanticAI to see what they're up to, since that's usually
| reflective of what's happening in broader industry.
|
| I'm not on X anymore so I can't follow many of the folks
| directly, but some of them (like Simon Willison) also post on
| BlueSky and Mastodon, fortunately. Some folks like Sebastian
| Raschka and Chip Huyen also post on LinkedIn. Kind of all over,
| but eventually, I see a good amount of what's happening.
| jameslk wrote:
| Maybe I've been missing some important stuff, but it seems the
| most relevant and important updates eventually just bubble up to
| the front page of HN or get mentioned in the comments
| mettamage wrote:
| This is my attitude for all tech surrounding IT
| SoftTalker wrote:
| Trying to keep up is like jumping on a 90mph treadmill. I have
| decided to opt out. I think AI (and currently LLMs) is more
| than a fad and not going away but it's in a huge state of churn
| right now. I'm not investing a ton of time into anything until
| I have to. In another few years the landscape will hopefully be
| more clear. Or not, but at least I won't have spent a lot of
| time on stuff that has quickly become irrelevant.
|
| I'm currently not using AI or LLMs in any of my day-to-day
| work.
| HellDunkel wrote:
| This. When has early adoptation paid off lately? Remember
| prompt engineering?
| Disposal8433 wrote:
| I remember the same FOMO that happened with "Google
| searching." People wrote and sold books about how to search
| properly on Google. The same with AI will happen: either it
| will flop, or we'll be able to learn all the needed skills
| in a few days. I don't worry about it.
| gowld wrote:
| I've banked professional success on my superior "Google
| Searching" ability (compared to my peers) for 20 years. A
| little but of thinking goes a long way, even when using
| tools designed to replace thought.
| Mars008 wrote:
| For some of us this time has come, i.e. it's already
| useful and worth learning and spending on prime accounts.
| twelve40 wrote:
| what do you mean remember? it didn't go anywhere. I try to
| understand how to make this useful for my daily
| programming, and every credible-looking advice begins with
| "tell LLM to program in style ABC and avoid antipatterns
| like XYZ", sometimes pages and pages long. It seems like
| without this prompt sourcery you cannot produce good code
| using an LLM it will make the same stupid mistakes over and
| over unless you try to pre-empt them with a carefully
| engineered upfront prompt. Aside from stupid "influencers"
| who bullshit that they produced a live commercial app with
| a one-liner English sentence, it seems that getting
| anything useful really requires a lot of prompt work,
| whatever you want to call it.
| bluefirebrand wrote:
| I agree pretty strongly
|
| Yeah it's not a fad, but I think it's really not as useful to
| me right now as the hype seems to suggest
|
| I'm going to keep an eye on developments, but I'm not using
| it for my day to day either. I'm just not seeing the same
| value other people seem to be seeing right now and I'm not
| going to exhaust myself trying to keep up
|
| One day Claude is the best. Next is Cursor. People are
| switching tools every two weeks trying to keep up
|
| Not for me thanks. Especially not with how inconsistent the
| quality and output still are
| bzmrgonz wrote:
| give me whisper real-time transcriber and I'll be happy!!
| chasd00 wrote:
| i'm in the same boat, i did the zero-to-hero series karpathy
| put out to get a basic understanding of what we're dealing
| with. At work, I helped put together a basic rag setup last
| year that's in production so feel ok with implementation.
| Finally, i have a little personal pet project that i'm going
| to feel out with claude code over Christmas to get more hands
| on. That's about it all i'm going to do to "keep up". I'll
| let the fighters fight it out and then just use whatever
| wins.
|
| btw here's a link to the karpathy videos
| https://karpathy.ai/zero-to-hero.html
|
| edit: i use claude and chatgpt day to day to help with simple
| things like regex, a replacement for google search in same
| cases, and self contained classes, functions, and other small
| discreet blocks of code.
| roboror wrote:
| Yeah these companies have made unbelievable investments,
| keeping their products a secret is antithetical to their
| existence.
| kybernetikos wrote:
| The big question for me at the moment is whether to go pay-for-
| tool or pay-for-model.
|
| If I pay for a tool that includes access to frontier models, then
| they'll keep the models up to date over time for me, let me use
| models from multiple providers, and the tool is carefully
| designed around the capabilities and limitations of the models it
| works with. On the other hand I can't really use the powerful
| model the tool works with for other applications or write my own.
|
| If I pay for models, then I can only really use it with that
| manufacturers tools or tools that aren't optimised for the model
| but allow you to bring your own keys, and if the model provider
| I'm paying falls behind then I'm tied in for the duration of the
| contract. The big advantage is that there is a lot of innovation
| in tooling happening at the moment and you can avoid being locked
| out of that or having to pay many times for access to the same
| frontier models accross multiple different tools.
| lumost wrote:
| I think we'll see a shift to BYO LLM in the future. I saw that
| overleaf offers an assistant feature and an associated
| subscription. But I already pay for ChatGPT and Github copilot,
| I'm not going to pay for a third assistant - particularly when
| I can just bypass overleaf and use copilot in vscode directly.
| kybernetikos wrote:
| I think commercially there'll be a really strong pull for dev
| tools to resell access to frontier models.
|
| It's hard to make money out of dev tools, but if you tie in a
| service people are prepared to pay hundreds of dollars a
| month for, then suddenly it looks easier to make money out of
| an otherwise unsaleable IDE plugin that accesses that
| service.
| vunderba wrote:
| Yeah, Raycast and Warp are two tools that are definitely
| trying to do this.
| smarx007 wrote:
| I think Github already offers BYOK (bring your own key). I
| think the problem is that you are still expected to pay for
| all the premium subs, the key only allows you to go above the
| rate/monthly limits.
|
| I think Jetbrains does it better with a full BYOM for models,
| including ollama. And I think if you go for ollama, you only
| need to pay for the IDE license, not for the AI add-on but
| don't quote me on that.
| vunderba wrote:
| For me the issue with paying for a tool (for example cursor) is
| that our incentives are not necessarily aligned.
|
| Since I'm not bringing my own API key, it's in their best
| interest to either throttle my usage by slowing it down or
| subtly downgrading me to a smaller LLM behind the scenes.
| twapi wrote:
| Simon Willison's weblog is almost enough, highest S/N
| desertmonad wrote:
| ^ https://simonwillison.net
| baal80spam wrote:
| And thankfully, he has an RSS!
| reactordev wrote:
| Andrej's talks have helped me tremendously. They're up on YT [0].
| For a long time I used to mentor and help machine learning
| scientists but when I hear Andrej speak, it's like I'm the
| student without any knowledge. It was a really strange feeling at
| first but I've come to value so much. I'm Jon Snow. I know
| Nothing (compared to Andrej).
|
| [0] https://www.youtube.com/@AndrejKarpathy
| n8cpdx wrote:
| Article does not make the case for why you must keep up.
|
| AFAICT the best strategy would have been to completely tune out
| AI for the last ~3 years:
|
| - AI has not meaningfully improved productivity (unless you're
| doing something super basic like react and were already bad at
| it). If you are using AI in a transformative way, that looks
| different today than it did 6 months ago. - AI has not stolen
| jobs (end of ZIRP did that) - The field changes so fast that you
| could completely tune out, and at any moment become up-to-date
| because the news from 3 months ago is irrelevant.
|
| I don't get where this meme that "you have to keep up" comes
| from.
|
| You have agency. You can get off the treadmill. You will be fine.
| bluefirebrand wrote:
| I have been tuning out for the last ~3 years and unfortunately
| it hasn't been the best strategy because the hype is still
| running roughshod all over me
|
| It is very likely my employer will use my AI apathy as an
| excuse to include me in the next round of layoffs, compared to
| my coworkers that are very AI enthusiastic
| scellus wrote:
| Maybe that would even be justified, if you are a software
| developer? If not now, at least soon.
|
| Imagine a software developer who refuses to use IDEs, or any
| kind of editor beyond sed, or version control, or some other
| essential tool. AI is soon similar, except in rare niche
| cases.
| sshine wrote:
| I've met many people smarter than myself who object to all
| of those.
|
| One professor thought syntax highlighting was a
| distraction.
|
| Lots of colleagues used vim/helix instead of IDEs.
|
| I haven't met anyone who refused version control from an
| intelligent standpoint.
|
| The most reasonable objection to AI from people who don't
| hate it are: I just don't know how it could
| help me; it's not as skilled as me at my job, and I'm
| already doing fine.
| hagbarth wrote:
| Version control is different since it's collaboration with
| the rest of the org.
|
| The rest: if they are just as productive as others, I would
| not care one bit. Tool use as a metric is just bad.
| chasd00 wrote:
| They're not that hard to learn how to use. If your employer
| asks you to use one then just learn whatever tool the license
| you for, it's not that big of a deal. It's like learning how
| to use your employer's official IDE or email client.
| sshine wrote:
| The article uses "why" in the title, but does not follow
| through with an answer.
|
| It sort of hints at one reason:
|
| > _The most common errors of misunderstanding are either
| underestimation ("it's all hype that will blow over") or
| overestimation ("I don't need programmers anymore"). These
| patterns are rooted in a lack of a solid understanding of the
| technology and how it is evolving over time._
|
| So if you don't at least find some middle ground between those
| two poles, you will make uninformed choices.
|
| But I agree: It is safe to ignore AI for now.
|
| I do sense that some people attach to AI because of a
| fundamental anxiety that it might transform society quickly and
| detrimentally, because that's part of the hype speech ("it will
| murder us all, it will make us all unemployed, it will turn us
| into slaves, maybe you can ride the dragon, and maybe you
| must").
|
| ---
|
| > _AI has not meaningfully improved productivity_
|
| This is contended.
|
| As the article says, we are in one of the most polluted
| information environments.
|
| People will say "It's absolutely useless" and "It has
| fundamentally changed my life."
|
| So neither extreme can be taken at face value as
| representative; they're samples of a murky picture.
|
| > _The field changes so fast that you could completely tune
| out_
|
| It's not that fast, in my opinion. Last big steps:
| - Transformer architecture (2017) - Larger models with
| greater performance (2020-) - Chain of thought (research
| in 2022, commercial breakthrough in 2024) - Agents (since
| forever, but 2022 for GPT-based agentic frameworks)
|
| Other things happened; for example, DeepSeek making an
| architectural breakthrough and challenging the financial model
| of open/closed weights.
|
| But most of the hype is just people trying to make commercial
| success on a few cornerstone breakthroughs.
|
| In one to two years, maybe we can add one more major
| advancement.
| lsy wrote:
| If you have a decent understanding of how LLMs work (you put in
| basically every piece of text you can find, get a statistical
| machine that models text really well, then use contractors to
| train it to model text in conversational form), then you probably
| don't need to consume a big diet of ongoing output from PR
| people, bloggers, thought leaders, and internet rationalists.
| That seems likely to get you going down some millenarian path
| that's not helpful.
|
| Despite the feeling that it's a fast-moving field, most of the
| differences in actual models over the last years are in degree
| and not kind, and the majority of ongoing work is in tooling and
| integrations, which you can probably keep up with as it seems
| useful for your work. Remembering that it's a model of text and
| is ungrounded goes a long way to discerning what kinds of work
| it's useful for (where verification of output is either
| straightforward or unnecessary), and what kinds of work it's not
| useful for.
| thorum wrote:
| Beyond a basic understanding of how LLMs work, I find most LLM
| news fits into one of these categories:
|
| - Someone made a slightly different tool for using LLMs (may or
| may not be useful depending on whether existing tools meet your
| needs)
|
| - Someone made a model that is incrementally better at
| something, beating the previous state-of-the-art by a few %
| points on one benchmark or another (interesting to keep an eye
| on, but remember that this happens all the time and this new
| model will be outdated in a few months - probably no one will
| care about Kimi-K2 or GPT 4.1 by next January)
|
| I think most people can comfortably ignore that kind of news
| and it wouldn't matter.
|
| On the other hand, some LLM news is:
|
| - Someone figured out how to give a model entirely new
| capabilities.
|
| Examples: RL and chain of thought. Coding agents that actually
| sort of work now. Computer Use. True end-to-end multimodal
| modals. Intelligent tool use.
|
| Most people probably _should_ be paying attention to those
| developments (and trying to look forward to what's coming
| next). But the big capability leaps are rare and exciting
| enough that a cursory skim of HN posts with >500 points should
| keep you up-to-date.
|
| I'd argue that, as with other tech skills, the best way to
| develop your understanding of LLMs and their capabilities is
| not through blogs or videos etc. It's to build something.
| Experience for yourself what the tools are capable of, what
| does and doesn't work, what is directly useful to your own
| work, etc.
| PaulHoule wrote:
| _Rewritten in response to quality complaint._
|
| A lot of people are feeling HN is saturated with AI posts
| whether it is how MCP is like USB-C (repeated so much you
| _know_ it is NPCs) or how outraged people are that their sh1t
| fanfics are being hoovered up to train AI.
|
| This piece is not "news", it's a summary which is tepid at
| best, I wish people had some better judgement about what they
| vote up.
| Fraterkes wrote:
| Just a heads up: you should try to get better at writing.
| PaulHoule wrote:
| Thanks for the tip!
| qsort wrote:
| I agree, but with the caveat that it's probably a bad time to
| fall asleep at the wheel. I'm very much a "nothing ever
| happens" kind of guy, but I see a lot of people who aren't
| taking the time to actually understand how LLMs work, and I
| think that's a huge mistake.
|
| Last week I showed some colleagues how to do some basic things
| with Claude Code and they were like "wow, I didn't even know
| this existed". Bro, what are you even doing.
|
| There is definitely a lot of hype and the lunatics on Linkedin
| are having a blast, but to put it mildly I don't think it's a
| bad investment to experiment a bit with what's possible with
| the SOTA.
| chamomeal wrote:
| I mean I didn't find out about Claude code until like a week
| ago and it hasn't materially changed my work, or even how I
| interact with LLMs. I still basically copy paste into claude
| on web most of the time.
|
| It is ridiculously cool, but I think anybody developer who is
| out of the loop could easily get back _into_ the loop at any
| moment without having to stay caught up most of the time.
| qsort wrote:
| I'm not talking about tools in particular, I completely
| agree that they're basically fungible, and for "serious"
| stuff it's probably still better to use the web interface
| directly as you have more control over the context.
|
| The problem I see is that a lot of people are grossly
| misaligned with the state of the art, and it does take a
| bit of experimentation to understand how to work with an
| LLM. Even basic stuff like how to work with context isn't
| immediately obvious.
| Fraterkes wrote:
| I don't think you're wrong, but if it takes someone a
| month (at most) to get up to speed with these tools, I
| don't think thats much of an argument for closely keeping
| up with them (until you need to know them to keep your
| job or smt) especially because everything is changing
| every few months. There is arguably no technology that
| needs you to "keep up" with it less
| crystal_revenge wrote:
| > I see a lot of people who aren't taking the time to
| actually understand how LLMs work
|
| The trouble is that the advice in the post will have very
| little impact on "understanding how LLMs work". The number of
| people who talk about LLMs daily but have never run an LLM
| local, and certainly never "opened it up to mess around" is
| very large.
|
| A fun weekend exercise that _anyone_ can do is to implement
| speculative decoding[0] using local LLMs. You 'll learn _a
| lot_ more about how LLMs work than reading every blog
| /twitter stream mentioned there.
|
| 0. https://research.google/blog/looking-back-at-speculative-
| dec...
| layer8 wrote:
| > the lunatics on Linkedin are having a blast
|
| That's a nice way to put it, made me chuckle. :)
| crystal_revenge wrote:
| I strongly agree with this sentiment and found the blog's list
| of "high signal" to be more a list of "self-promoting" (some
| good people who I've interacted with a fair bit on there, but
| that list is more 'buzz' than insight).
|
| I also have not experienced the post's claim that: "Generative
| AI has been the fastest moving technology I have seen in my
| lifetime." I can't speak for the author, but I've been in this
| field from when "SVMs are the new hotness and neural networks
| are a joke!" to the entire explosion of deep learning, and
| insane number of DL frameworks around the 20-teens, all within
| a decade (remember implementing restricted Boltzmann machines
| and pre-training?). Similarly I saw "don't use JS for anything
| other than enhancing the UX" to single page webapps being the
| standard in the same timeframe.
|
| Unless someone's aim is to be on that list of "High signal"
| people, it's far better to just keep your head down until you
| actually need these solutions. As an example, I left webdev
| work around the time of backbone.js, one of the first attempts
| at front end MVC for single pages apps. Then the great
| React/Angular wars began, and I just ignored it. A decade later
| I was working with a webdev team and learned React in a few
| days, very glad I did not stress about "keeping up" during the
| period of non-stop changing. Another example is just 5 years
| ago everyone was trying to learn how to implement LSTMs from
| scratch... only to have that model essentially become obsolete
| with the rise of transformers.
|
| Multiple times over my career I've learned lesson that _moving
| fast_ is another way of saying _immature_. One would find more
| success learning about the GLM (or god forbid understanding to
| identify survival analysis problems) and all of it 's still
| under appreciated uses for day-to-day problem solving (old does
| not imply obsolete) than learning the "prompt hack of the
| week".
| alphazard wrote:
| When explaining LLMs to people, often the high level
| architecture is what they find the most interesting. Not the
| transformer, but the token by token prediction strategy
| (autoregression), and not always choosing the most likely
| token, but a token proportional to its likelihood.
|
| The minutiae of how next token prediction works is rarely
| appreciated by lay people. They don't care about dot products,
| or embeddings, or any of it. There's basically no advantage to
| explaining how that part works since most people won't
| understand, retain, or appreciate it.
| panarchy wrote:
| AI research was so interesting pre-transformers it was starting
| to get a bit wild around GPT2 IIRC but now the signal to noise
| is so low with every internet sensationalist and dumb MBA
| jumping on the bandwagon.
| helloplanets wrote:
| It's not a model of text, though. It's a model of multiple
| types of data. Pretty much all modern models are multimodal.
| victorbjorklund wrote:
| Indeed. We have just had a few really big shifts since launch
| of GPT3. Rest has just been bigger and more optimized models +
| tooling around the models.
| blactuary wrote:
| Actually no I do not have to keep up
| paul7986 wrote:
| Indeed just get out of tech and make a new living! Tech jobs
| are declining and will continue then fall off a cliff with one
| doing the job ten use to. Followed by other white collar and
| blue collar (AMazons warehouse robots) jobs.
|
| Happily canceled my GPT Plus this week; personally not gonna
| feed that beast any longer! As well it can not generate maps
| (create road trip travel maps showing distance between
| locations to share with friends, a creek tubing map route &
| etc) at all like Gemini can for free.
| astrange wrote:
| > Tech jobs are declining and will continue then fall off a
| cliff with one doing the job ten use to.
|
| This would increase employment ceteris paribus. That's like
| saying inventing new programming languages is bad because
| they're too productive.
|
| https://en.wikipedia.org/wiki/Jevons_paradox
| kybernetikos wrote:
| >> one doing the job ten use to.
|
| > This would increase employment ceteris paribus.
|
| This might be true, but if it is, the one "doing the job
| ten use[d] to" would not actually being doing the same kind
| of work at all, and so therefore might not be the same
| people or even same kind of people. Even if we do Jevons
| ourselves out of this situation, it might still spell
| employment disaster for mid level coders, while increasing
| employment for product managers.
| astrange wrote:
| Indeed that's the case, but it's sort of what software
| engineers signed up for anyway. I don't know if people
| are expecting to go their whole careers without learning
| a new stack / language. Maybe they are?
| dingnuts wrote:
| That's right, there are no carpenters or lumberjacks anymore
| because power tools were invented
| ninetyninenine wrote:
| Obviously AI is different. While LLMs are more of a power
| tool now, the future trendline points towards something
| that (possibly) replaces us. That's the entire concern
| right? I mean everyone knows this.
|
| Is this not obvious?
|
| Why do people hide behind this ridiculous analogy: "That's
| right, there are no carpenters or lumberjacks anymore
| because power tools were invented"
|
| ???
|
| I mean sure the analogy is catchy and makes surface level
| sense, but can your brain do some analysis outside the
| context of an analogy??? It makes no sense that all of AI
| can be completely characterized by an analogy that isn't
| even accurate yet people delusionally just regurgitate the
| analogy most fitting with the fantasy reality they prefer.
| lurking_swe wrote:
| sure. But why is digging one's head in the sand a good
| strategy? To be clear i'm not advocating for trying to
| keep up all the time. You gotta live life too. But
| "ejecting" completely is dumb.
|
| Are you saying that when the sewing machine was invented,
| it would be in the employees interest to not learn how to
| use it? Or when the computer was invented, it's not in
| the employees interest to learn how to use it?
|
| Even if you are a software engineer and are fired / laid
| off / pushed out of the industry because of AI, knowing
| how to use AI, its risks, etc is still helpful. It's a
| skill you can hopefully use in your next career, whatever
| you pivot to. Unless you pivot to manual labor.
|
| Thinking otherwise is shortsighted.
| kybernetikos wrote:
| There are lots of careers that used to be widespread but
| are now extremely niche because of technological change.
| For example, most careers relating to horses.
| paul7986 wrote:
| Exactly hence why i said its probably wise to not pursue
| tech unless your an AI researcher PhD level already,
| striving for such in school studies and or leveling up
| your skills in AI.
| paul7986 wrote:
| So say you are a startup are you going to now hire a
| designer or use WAY Less Expensive & quicker AI to design
| logos, website, an app, etc? Print design.. all types of
| design it can do.
|
| So You and all other people like to save money are going to
| continue spend the same thousands on such a resource when
| AI can do what they do in a few minutes or more for WAY
| LESS? UX Design was supposedly a growing field ... not at
| anymore! Definitely one can do the same thing in that field
| that 10 did.
|
| Further, future mobile AI devices will pull the information
| and put it all on the lock screen of your AI device
| visualizing the data in a fun new way. Technology that
| makes things simpler and more magical get adopted yet
| visits to websites will significantly decline.
|
| For federal workers who have lost their jobs they are
| feeling this pain competing for jobs against each other and
| now AI. It will only get worse for designers because it's
| now cheaper and faster to use AI to design logos, sites,
| apps to even including do vibe coding for the front end
| development to possibly the backend but that's not my
| specialty yet no doubt I vibe coded front-ends.
| mertleee wrote:
| Yeah. Time to do something else.
| phyzome wrote:
| The article doesn't actually explain the "why", which severely
| undercuts the existence of the "how" list.
|
| It's fine to go do other things with your precious time instead.
| pizzathyme wrote:
| Exactly. I have been questioning the need to "keep up". As the
| best AI innovations arrive, the ones that are useful will
| eventually make their way to the mainstream (and me). I didn't
| "keep up" with the nascent development of spreadsheets or
| Google Docs. Once they were widely used, I adopted them just
| fine, and haven't missed out on anything.
|
| Unless you are building an AI startup furiously searching for
| PMF before your runway expires, I don't see the urgency.
| gowld wrote:
| Someone who keeps up on AI and finds productivity gains will
| outcompete someone who doesn't, even for activities that
| aren't develoing new AI weren't AI-based before.
|
| What is an "AI startup"? If you add a chatbot do your
| product, are you an "AI startup"? Does "startup" require
| having no moat? Can you be an established stable business
| that loses everything to a startup that leapfrogs your tech,
| AltaVista?
| codebolt wrote:
| Just subscribing to OpenAI, Anthropic and Google on YouTube is
| pretty helpful. They post demos of all major new feature releases
| that are good to skim through to get a sense of where the
| frontier is moving (just take all claims about capabilities with
| a grain of salt).
|
| I've also got some gems from Microsofts Build talks, specifically
| whenever Scott Hanselman and Mark Russinovich get together, e.g.:
| https://www.youtube.com/watch?v=KIFDVOXMNDc
| linsomniac wrote:
| Here's my current rule of thumb: If you have successfully built a
| couple projects using agentic tooling and Claude 4 or similar
| models: you are doing a fine job of keeping up. Otherwise, you
| are at least a generation behind.
| bluefirebrand wrote:
| Behind what?
|
| Isn't the whole promise of AI tools that they just work?
|
| What skill am I missing out on learning, exactly, by not using
| them right now? Prompt Engineering?
|
| I think I'm a reasonably good communicator in both voice and
| text, so what skill am I failing to train by not using LLMs
| right now?
| linsomniac wrote:
| >Isn't the whole promise of AI tools that they just work?
|
| No, not at all. Like with pretty much any development tool,
| you need to get proficient with them.
|
| >what skill am I missing out on
|
| At this point, it seems like pretty much all of them related
| to generative AI. But, the most recent of them that I'll
| point at is: tooling tooling tooling, and prompting. But the
| specific answer (to answer your "exactly") is going to depend
| on you and what problems you are solving. That's why on tries
| not to fall behind, so you can see how to use tooling in a
| rapidly evolving landscape, for your exact circumstances.
|
| >I think I'm a reasonably good communicator in both voice and
| text, so what skill am I failing to train by not using LLMs
| right now?
|
| You know how to achieve something you will use different
| words with different people? You don't talk to your spouse
| the same way you talk to your parents or your children or
| your friends or your coworkers, right? You understand that if
| you are familiar with someone you speak to them differently
| if you want to achieve something, yes?
| dingnuts wrote:
| this is just ridiculous. you can get up to speed with SOTA
| tooling in a few hours. A system prompt is just a prompt
| that runs every time. Tool calls are just patterns that are
| fine tuned into place so that we can parse specific types
| of LLM output with traditional software. Agents are just a
| LLM REPL with a context specific system prompt, and limited
| ability to execute commands
|
| none of this stuff is complicated, and the models
| themselves have been basically the same since GPT-2 was
| released years ago
| chasd00 wrote:
| > A system prompt is just a prompt that runs every time.
| Tool calls are just patterns that are fine tuned into
| place so that we can parse specific types of LLM output
| with traditional software. Agents are just a LLM REPL
| with a context specific system prompt, and limited
| ability to execute commands
|
| pulling the covers back so hard and so fast is going to
| be shocking for some.
|
| To make it more concrete you can try and build something
| yourself. Grab a small model off of hugging face that you
| can run locally. Then put a rest API in front of it so
| you can make a request with curl, send in some text, and
| get back in the response what the llm returned. Now in
| the API prepend some text to what came on the request (
| this is your system prompt ) like "you are an expert
| programmer, be brief and concise when answering the
| following". Now add a session to your API and include the
| past 5 requests from the same user along with the new one
| when passing to the llm. Update your prepended text (the
| system prompt) with "consider the first 5
| requests/responses when formulating your response to the
| question". you can see where this is going, all of the
| tools and agents are some combination of the above and/or
| even adding more than one model.
|
| At the end of the day, everyone has a LLM at the core
| predicting and outputting the next most likely string of
| characters that would follow from an input string of
| characters.
| Disposal8433 wrote:
| Your FOMO threats are way too obvious, and I'm not falling for
| it . I'm behind you in C++ because I don't use AI? That's
| ridiculous.
| linsomniac wrote:
| It may not be as ridiculous as you think. I have 25 years of
| experience with Python, and the generative AI tooling is
| teaching me useful things in Python when I work with it.
| ethan_smith wrote:
| Building projects is valuable, but "keeping up" is contextual -
| someone using AI effectively in their specific domain is ahead
| regardless of which generation of tools they're using.
| gtsop wrote:
| Right now they only thing being left behind is my actual work
| to be done since I am spending more and mpre time fighting off
| cursor-written degenerate slop code from creeping into the
| codebase from "pioneer" developers who are slowly forgetting
| how to program.
| nice_byte wrote:
| Doesn't explain the "why you must too" part
| tezza wrote:
| Add to the list:
|
| https://x.com/rowancheung - Rowan Cheung: "Daily" updates and
| insider access
| umanwizard wrote:
| > and why you must too
|
| No, I don't think I do. Been working great for me so far.
| deadbabe wrote:
| Don't you _ever_ let people make you think you have to tirelessly
| follow some tech progress or stay on top of things all the time.
|
| Progress is like a bus. You can just get on board at any time.
| You're not going to fall behind. And staying up to date doesn't
| keep you "ahead" of anyone.
|
| Doing things is what gets you ahead, and if you don't feel like
| doing something right now, don't worry about, do something later,
| and you'll be ahead of people who aren't doing anything at that
| moment.
| scellus wrote:
| If one wants to follow AI development mostly in the sense of LLMs
| and associated frontier models, that's an excellent list with
| over half of the names familiar, to whom I have converged
| independently.
|
| I have a list in X for AI; it's the best source of information
| overall on the subject, although some podcasts or RSS feeds
| directly from the long-form writers would be quite close. (If one
| is a researcher themselves, then of course it's a must to follow
| the paper feeds, not commentary or secondary references.)
|
| I'd add https://epoch.ai to the list, on podcasts at least
| Dwarkesh Patel; on blogs Peter Wildeford (a superforecaster),
| @omarsar0 aka elvis from DAIR in X, also many researchers
| directly although some of them like roon or @tszzl are more
| entertaining than informative.
|
| The point about polluted information environment resonates on me;
| in general but especially with AI. You get a very incomplete and
| strange understanding by following something like NYT who seem to
| concentrate more on politics than technology itself.
|
| Of course there are adjacent areas of ML or AI where the sources
| would be completely different, say protein or genomics models, or
| weather models, or research on diffusion, image generation etc.
| The field is nowadays so large and active that it's hard to grasp
| everything that is happening on the surface level.
|
| Do you _have_ to follow? Of course not, people over here are just
| typically curious and willing to follow groundbreaking
| technological advancements. In some cases like in software
| development I'd also say just skipping AI is destructive to the
| career in the long term, although there one can take a tools
| approach instead of trying to keep track of every announcement.
| (My work is such that I'm expected to keep track of the whole
| thing on a general level.)
| rglover wrote:
| You don't need to "keep up," you just need to loosely pay
| attention to identify things/features that will make you more
| productive, test them out, and keep what actually works (not what
| some influencer claims to work on social media). In fact, I feel
| far more confident in my understanding by listening to
| researchers who _dismiss_ the wild claims about AI 's potential--
| not hype it blindly [1].
|
| There's far too much noise, churn, and indecision at this stage
| to get any productive value out of riding the bleeding edge.
|
| If it's actually revolutionary, you'll hear about it on HN.
|
| [1] https://x.com/burkov
___________________________________________________________________
(page generated 2025-07-18 23:00 UTC)