Transcript
Screenshot of a Mastodon post by Kevin Beaumont: “Generative Al government lobbying.”
Photo of AI/tech company CEO’s, captioned:
We spent a Trillion on NVDA GPUs antide dont have any Al product you want.
Photo of a crying male, captioned:
Please like our Al bro This is the last time bro. So many possibilities bro. Its the future bro. Just need you to like it bro. We worked real hard bro. Our stockholders need this one bro.
Lemmy is losing it’s collective mind over AI. It’s so bizarre to witness such a willful rejection of reality. I was able to understand some really poorly documented code just today in a few minutes using it instead of hours it would take to dig through it. Is it worth the collective cost in energy and investment to build it in the first place? Who knows, that would be the rational conversation to have. But Lemmy is in full denial on this one
There’s something about certain people in this community and I don’t know what it is but they don’t seem to be able to understand the most fundamental point of all which is that 99% of the population of the planet aren’t coders.
AI is useless to me because I don’t really spend much time programming.
Reverse dunning-kruger does play out often here.
Recently i said out loud “Well when most people hear ‘semiconductors’ they automatically think ‘silicon’, however…” and then trailed off as I realized my audience (parents asking me what I do at work) they had immediately stopped listening.
This plus woosh is lemmy 🤣
Fine, another anecdote, I know several people that don’t use search engines anymore and go directly to chatgpt. Do I think that’s good? No but I’m not pretending it’s not happening
Until it can ingest arbitrary URLs that contain documentation, images that contain relevant information need to be OCR into knowledge articles, random emails somebody sent me 6 months ago, and teams chats, it isn’t much used to me.
The truth is most work that AI could do hasn’t been optimised for AI usage it’s been optimised towards random organic human usage which AI is totally incapable of ingesting.
Please come back to me when it can deal with the six different VPNs I have to manage in order to be barely competent to my job
Oh and of course the timesheet application which apparently is super duper important.
The real world is messy, inconsistent, and requires planning to document.
But you tell me about all your friends that swear by it, I’m sure their experiences relevant to me.
Wait so I give a technical anecdote and then a non technical anecdote and now I’m told the non technical one doesn’t pertain to your technical application? You doing okay over there?
It’s not a technical anecdote.
It’s just an example of the problem with the technology. Why do you presume that the end result of all of that is anything remotely technical? The end result of all of that is a really complicated and boring subject matter to mostly do with how you build roads, it involves computers but the end result isn’t computers.
Your problem is you have a certain idea of what AI is and what it can be used for and you forget that that doesn’t apply to the whole universe.
This is a technical anecdote. What is wrong with you?
To each their own… I have tried using AI to code simple stuff to just give up in an hour as I gauge it would have taken me less than that to just code it myself.
At the moment, the best I can use it for is to remind me of syntax, get me some references and/or produce some fake data to test. To that I say, it is useful, just not worth the money or the harm to the planet… much less giving any more of our data and privacy to the tech bros
Technology Example Maybe Useful For Marketed As Actual Use ---------- ------- ---------------- ------------- ------------- Voice Recognition Alexa Voice Typing OMG!!! Digital Assistant!!!!! Hoover up your personal information LLM ChatGPT Better Google OMG!!! Replace people!!!! Hoover up your personal information Blockchain - ? OMG!!!! Put in everything!!!!! Cryptocurrency NPT - N/A OMG!!!! NPT!!!! Fleecing rubes CryptoCurrency BitCoin Online $ transfer OMG!!!! Buy! Buy! Buy! Stealing your real moneyN.B. There has GOT to be a better way to align tables in Lemmy.
LLMs are definitely not for better Google. They are text generators, not search engines.
They could be used in conjunction with search engines, like perplexity does. But they can not and will not ever replace them.
They make tons of mistakes and they can not be fully trusted. So don’t use them as truth or things that require to be faultless.
LLMs are useful for descriptions (like comments in code), creative input (help writers get out of writers block), summaries (full text can be checked for hallucinations), and corporate emails (since nobody reads that flowerly language anyway).
Nothing is worth giving any corporate entity control of your data or privacy. Not even ASI
You will find some lemmings that hate ‘AI’ for correct and nuanced reasons. Not in a meme community post probably since I myself take licence to just be a snarky jerk in these type of communities. I come here to make fun of tech bros and such. They definitely deserve it.
There are a lot of highly skilled people around and you can find lots of opportunities for really interesting discussions about the proper uses of ‘AI’. Unlike blockchain which has turned out to have very niche applications beyond the dubious utility of crypto, these things definitely have real power to accelerate progress in many directions.
But come on, can’t we just rag on the hugely visible dipshittery that is accompanying this obvious ‘tuplip mania’?
To be fair, which application of AI has been more useful than “save money without government”?
I’m all for making fun of tech but if the punchline is just inaccurate it loses me
:.|:;Losses me
If AI can do everything for you, then what are you needed as a programmer? You don’t seem to understand that magic wands in our world have their own huge price.
Who said AI can do everything for you??? And regardless, how does pretending it’s totally useless serve me? You think I keep my job if I just put my head in the sand? I think you totally misunderstand what I wrote
How can I express myself more precisely, because I often express my thoughts poorly due to fatigue… This was an example, not a clear statement.
But even as an example, it misses my point. Pretending I still have a job in that hypothetical scenario does not help me when I lose my job. Being aware of it gives me time to plan and respond
Edit: ok seeing as I keep getting downvoted and you’re getting upvoted, either my original belief that Lemmy is losing it’s mind is correct or I’m missing something about what you’re saying? Can you elaborate?
Post making a point: 24 points
Reply sharing a personal experience supporting the point: -13 points
lol
I suspect a large chunk of people in denial have either never used it and are basing their judgment on memes and cherry-picked failure modes that go viral, or have only used the earlier/weaker models that are far more prone to hallucination. I use ChatGPT Plus regularly, and find it occasionally imperfect but still very useful, but a few times I’ve tried using it while logged out and the free model it defaults to was far more flawed. Probably why OpenAI built GPT5 around a marginally better but far more efficient model that they can roll out to everyone (including the large cohort of free users).
Well personally I have used it but I can’t really find anything useful for it to do. It’s very mathematics and programming orientated which is useful if your job involves a lot of mathematics or programming. If not erh, it’s basically just an interesting toy.
It’s bloody rubbish at civil engineering I’ll tell you that.
Why anyone thinks using a Large Language Model can do anything involving math is beyond me.
It is like using a calculator to write a book because it can spell 80085.
Use it for documentation, not for coding, maths, or engineering.
I had to parse a bunch of SQL, with which I’m not familiar, and while trying to understand an issue I was having I solved it. It’s a powerful tool that can guide you to understanding so well that you understand the problem and the solution at the same time. Or maybe I was just lucky.
The problem is when I use LLMs to analyze a problem in a domain I do have expertise in, I find that its interpretations are often very wrong and misleading. This is why I definitely don’t trust it in domains I lack expertise in.
In my experiences it’s a powerful tool that requires luck lol