Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

LLM Inevitabilism

tomrenner.com

Post reply on HN
Print view View original

811–820 of 1001 posts

Re: LLM Inevitabilism

#811

Post by dmbche »

xandrius wrote: post #804
bloppe wrote: post #781

This inevitabilist framing rests on an often unspoken assumption: that LLM's will decisively outperform human capabilities in myriad domains. If that assumption holds true, then the inevitabilist quotes featured in the article are convincing to me. If LLM's turn out to be less worthwhile at scale than many people assume, the inevitabilist interpretation is another dream of AI summer. Burying the core assumption and f…

LLMs have already been absolutely worthwhile in many of my projects, so I guess it's already inevitable for me.

>that LLM's will decisively outperform human capabilities in myriad domains.

Do your LLMs outperform you at your tasks?

If not, were they to become more expensive by a non negligible margin, would you at any cost keep using them in their curent state?

Re: LLM Inevitabilism

#812

Post by thewebguyd »

ohyes wrote: post #783
underdeserver wrote: post #740

Earlier quoted context omitted.

> That is a such a wild claim. People like the output of LLMs so much that ChatGPT is the fastest growing app ever. The people using ChatGPT like its output enough when they're the ones reading it. The people reading ChatGPT output that other people asked for generally don't like it. Especially if it's not disclosed up front.

Had someone put up a project plan for something that was not disclosed as LLM assisted output. While technically correct it came to the wrong conclusions about the best path forward and inevitably hamstrung the project. I only discovered this later when attempting to fix the mess and having my own chat with an LLM and getting mysteriously similar responses. The problem was that the assumptions made when asking the LL…

> do not have the ability to challenge your assumptions or think laterally.

Particularly on the challenging your assumptions part is where I think LLMs fail currently, though I won't pretend to know enough about how to even resolve that; but right now, I can put whatever nonsense I want into ChatGPT and it will happily go along telling me what a great idea that is. Even on the remote chance it does hint that I'm wrong, you can just prompt it into submission.

None of the for-profit AI companies are going to start letting their models tell users they're wrong out of fear of losing users (people generally don't like to be held accountable) but ironically I think it's critically important that LLMs start doing exactly that. But like you said, the LLM can't think so how can it determine what's incorrect or not, let alone if something is a bad idea or not.

Interesting problem space, for sure, but unleashing these tools to the masses with their current capabilities I think has done, and is going to continue to do more harm than good.

Re: LLM Inevitabilism

#813

Post by henryfjordan »

pera wrote: post #742
brokencode wrote: post #706

Earlier quoted context omitted.

> “most people agree that the output is trite and unpleasant to consume” That is a such a wild claim. People like the output of LLMs so much that ChatGPT is the fastest growing app ever. It and other AI apps like Perplexity are now beginning to challenge Google’s search dominance. Sure, probably not a lot of people would go out and buy a novel or collection of poetry written by ChatGPT. But that doesn’t mean the outp…

> People like the output of LLMs so much that ChatGPT is the fastest growing app ever While people seem to love the output of their own queries they seem to hate the output of other people's queries, so maybe what people actually love is to interact with chatbots. If people loved LLM outputs in general then Google, OpenAI and Anthropic would be in the business of producing and selling content.

Google does put AI output at the top of every search now, and sometimes it's helpful and sometimes it's crap. They have been trying since long before LLMs to not just provide the links for a search but also the content.

Google used to be interested in making sure you clicked either the paid link or the top link in the results, but for a few years now they'd prefer that a user doesn't even click a link after a search (at least to a non-Google site)

Re: LLM Inevitabilism

#814

Post by justfix17 »

ohyes wrote: post #783
underdeserver wrote: post #740

Earlier quoted context omitted.

> That is a such a wild claim. People like the output of LLMs so much that ChatGPT is the fastest growing app ever. The people using ChatGPT like its output enough when they're the ones reading it. The people reading ChatGPT output that other people asked for generally don't like it. Especially if it's not disclosed up front.

Had someone put up a project plan for something that was not disclosed as LLM assisted output. While technically correct it came to the wrong conclusions about the best path forward and inevitably hamstrung the project. I only discovered this later when attempting to fix the mess and having my own chat with an LLM and getting mysteriously similar responses. The problem was that the assumptions made when asking the LL…

> LLMs do not think

Yep.

More seriously, you described a great example of one of the challenges we haven't addressed. LLM output masquerades as thoughtful work products and wastes people's time (or worse tanks a project, hurts people, etc).

Now my job reviewing work is even harder because bad work has fewer warning signs to pick up on. Ugh.

I hope that your workplace developed a policy around LLM use that addressed the incident described. Unfortunately I think most places probably just ignore stuff like this in the faux scramble to "not be left behind".

Re: LLM Inevitabilism

#815

Post by atleastoptimal »

desumeku wrote: post #518
atleastoptimal wrote: post #265

AI is being framed as the future because it is the future. If you can't see the writing on the wall then you surely have your head in the sand or are seeking out information to confirm your beliefs. I've thought a lot about where this belief comes from, that belief being the general Hacker News skepticism towards AI and especially big tech's promotion and alignment with it in recent years. I believe it's due to fear…

>This increasingly likely future is made all the more infuriating by the annoyances of the current reality of AI. The fact that AI is so presently inescapable despite how many glaring security-affecting flaws it causes, how much it propagates slop in the information commons, and how effectively it emboldens a particularly irksome brand of overconfidence in the VC world is preemptive insult to injury in the lead up a…

How could you take what I said and conclude that I’m stating that AI is completely useless and a net negative? I bring those issues up to note that the rollout of AI systems have been mired in problems like all new technologies. However even current AI possesses enormous utility in many sectors, and productivity/efficiency gains due to capabilities of the best models, which have no signs of slowing their rate of improvement.

That you rushed to straw man my point so hyperbolically supports my view that this frequent insistence I see claiming that AI is a worthless scam despite all the evidence to the contrary is emotionally motivated.

Re: LLM Inevitabilism

#817

Post by sparky4pro »

This is a nice article as it triggers “think before you use it” mentality.

However, at the same time, it suggests the idea that rational thinking without any deep seated perception or hidden motivation is possible.

This is not possible.

Therefore, all greedy people in this field will push anything that gives them what they want.

They will never care if what they do or promote will help “mankind” to a long term beneficial direction.

Re: LLM Inevitabilism

#818

Post by DanMcInerney »

jdauriemma wrote: post #732
DanMcInerney wrote: post #712

These articles kill me. The reason LLMs (or next-gen AI architecture) is inevitably going to take over the world in one way or another is simple: recursive self-improvement. 3 years ago they could barely write a coherent poem and today they're performing at at least graduate student level across most tasks. As of today, AI is writing a significant chunk of the code around itself. Once AI crosses that threshold of con…

> they're performing at at least graduate student level across most tasks I strongly disagree with this characterization. I have yet to find an application that can reliably execute this prompt: "Find 90 minutes on my calendar in the next four weeks and book a table at my favorite Thai restaurant for two, outside if available." Forget "graduate-level work," that's stuff I actually want to engage with. What many peopl…

This is absolutely doable right now. Just hook claude code up with your calendar MCP server and any one of these restaurant/web browser MCP servers and it'll do this for you.

https://apify.com/canadesk/opentable/api/mcp https://github.com/BrowserMCP/mcp https://github.com/samwang0723/mcp-booking

Re: LLM Inevitabilism

#819

Post by dmbche »

imiric wrote: post #112
Animats wrote: post #46

There may be an "LLM Winter" as people discover that LLMs can't be trusted to do anything. Look for frantic efforts by companies to offload responsibility for LLM mistakes onto consumers. We've got to have something that has solid "I don't know" and "I don't know how to do this" outputs. We're starting to see reports of LLM usage having negative value for programmers, even though they think it's helping. Too much eff…

> Look for frantic efforts by companies to offload responsibility for LLM mistakes onto consumers. Not just by companies. We see this from enthusiastic consumers as well, on this very forum. Or it might just be astroturfing, it's hard to tell. The mantra is that in order to extract value from LLMs, the user must have a certain level of knowledge and skill of how to use them. "Prompt engineering", now reframed as "con…

Simple thought I had reading this:

I've used a tool to do a task today. I used a suction sandblasting machine to remove corrosion from a part.

Without the tool, had I wanted to remove the corrosion, I would've spent all day (if not more) scraping it with sandpaper (is that a tool too? With the skin of my hands then?) - this would have been tedious and could have taken me all day, scraping away millimeter by millimeter.

With the machine, it took me about 3 minutes. I necessitated 4-5 minutes of training to attain this level of expertise.

The worth of this machine is undeniable.

How is it that LLMs are not at all so undeniably efficient? I keep hearing people tell me how they will take everyones job, but it seems like the first faceplant from all the big tech companies.

(Maybe second after Meta's VR stuff)

Re: LLM Inevitabilism

#820

Post by brokencode »

pera wrote: post #742
brokencode wrote: post #706

Earlier quoted context omitted.

> “most people agree that the output is trite and unpleasant to consume” That is a such a wild claim. People like the output of LLMs so much that ChatGPT is the fastest growing app ever. It and other AI apps like Perplexity are now beginning to challenge Google’s search dominance. Sure, probably not a lot of people would go out and buy a novel or collection of poetry written by ChatGPT. But that doesn’t mean the outp…

> People like the output of LLMs so much that ChatGPT is the fastest growing app ever While people seem to love the output of their own queries they seem to hate the output of other people's queries, so maybe what people actually love is to interact with chatbots. If people loved LLM outputs in general then Google, OpenAI and Anthropic would be in the business of producing and selling content.

I think the thing people hate about that is the lack of effort and attention to detail. It’s an incredible enabler for laziness if misused.

If somebody writes a design or a report, you expect that they’ve put in the time and effort to make sure it is correct and well thought out.

If you then find the person actually just had ChatGPT generate it and didn’t put any effort into editing it and checking for correctness, then that is very infuriating.

They are essentially farming out the process of creating the document to AI and farming out the process of reviewing it to their colleagues. So what is their job then, exactly?

These are tools, not a replacement for human thought and work. Maybe someday we can just have ChatGPT serve as an engineer or a lawyer, but certainly not today.

Post reply on HN