Live data from Hacker News

Ask HN: Is anyone else bearish on OpenAI?

news.ycombinator.com

281–290 of 321 posts

Re: Ask HN: Is anyone else bearish on OpenAI?

#281

I think many people are really underestimating the "intelligence" of an LLM. People have this misconception that LLMs are just complicated Markoc chains, predicting text based purely on probability. They are not. As it turns out, to accurately predict text, you need to learn a lot about the world. In fact, there's lots of fascinating things hidden in the weights of LLMs, like a geographic model of the world. [1] To m…

Information compression is nothing new. Of course if you are trying to get an LLM to produce accurate data about geography, it would make sense that instead of trying to catalog every fact in a lookup table, there would be some sort of model that you can run queries through. The LLMs just happen to be a neat way of building those models automatically from data.

The problem is, this sentiment of path towards AGI assumes that everything (i.e our physical reality) can be compressed. Which is highly likely not true.

The future of AI is basically better and better compression with more data sources. You will be able to do things like ask how to build a flying car in your garage, and it will give you step by step instructions on what things you need to order from where, including CAD drawings to get custom made CNC parts, and how to put it all together (including software to run it).

As far as AGI goes, its possible that through AI assisted technology we will be able to measure synapses of a human brain in fine enough detail, or somehow mass surveil the entire population of earth and derive human brain models from that. And then, with optimizations on that, we can potentially arrive at the most efficient human brain in terms of some metric, and we will have very good robotic assistants, but nothing really beyond that.

Re: Ask HN: Is anyone else bearish on OpenAI?

#282

Earlier quoted context omitted.

> I think this speaks to the bubble you're in and how your kind of software development probably represents 0.1% of what's out there. Grandparent specifically said they find ChatGPT useful for things outside their wheelhouse: >> [...]troubleshooting situations that flow into me - working in areas that I have the slimmest of backgrounds in It sounds to me like it may not be safe to use ChatGPT if you can't check its w…

Not the original poster, but here are some examples: - Paste a bunch of log code about which I have no idea. Ask to identify and explain the problem. - Wireshark / dmesg / OpenWrt configuration pasting, ask to fix the problem. For instance, I fixed a Wi-Fi issue in a heterogeneous setup, which turned out to be caused by a stray DHCPv6 server. - Paste C code, along with an error log, and ask to fix the problem. - Past…

Yeah I think the people who haven't found value are people who either don't have strong communication skills hence cannot articulate what they need OR work in very speciailized fields - either way they're clearly the minority seeing how many paying users chatgtp has and how fast they got there

Re: Ask HN: Is anyone else bearish on OpenAI?

#283

Earlier quoted context omitted.

GPT-4 or GPT-3.5? They're completely different products.

We need to stop propagating this nonsense. GPT-4 still messes up.

It's not nonsense at all. The degree to which GPT-4 messes up is dramatically lower than 3.5.

Re: Ask HN: Is anyone else bearish on OpenAI?

#284
post #282

Earlier quoted context omitted.

Not the original poster, but here are some examples: - Paste a bunch of log code about which I have no idea. Ask to identify and explain the problem. - Wireshark / dmesg / OpenWrt configuration pasting, ask to fix the problem. For instance, I fixed a Wi-Fi issue in a heterogeneous setup, which turned out to be caused by a stray DHCPv6 server. - Paste C code, along with an error log, and ask to fix the problem. - Past…

Yeah I think the people who haven't found value are people who either don't have strong communication skills hence cannot articulate what they need OR work in very speciailized fields - either way they're clearly the minority seeing how many paying users chatgtp has and how fast they got there

> don't have strong communication skills hence cannot articulate

Basically this - even in a highly specialised field you would still find value in it if you're providing enough context and describing the problem properly.

People don't want to hear that they're poor communicators - it's much easier to blame the tech.

Re: Ask HN: Is anyone else bearish on OpenAI?

#285

Earlier quoted context omitted.

> I think this speaks to the bubble you're in and how your kind of software development probably represents 0.1% of what's out there. Grandparent specifically said they find ChatGPT useful for things outside their wheelhouse: >> [...]troubleshooting situations that flow into me - working in areas that I have the slimmest of backgrounds in It sounds to me like it may not be safe to use ChatGPT if you can't check its w…

Not the original poster, but here are some examples: - Paste a bunch of log code about which I have no idea. Ask to identify and explain the problem. - Wireshark / dmesg / OpenWrt configuration pasting, ask to fix the problem. For instance, I fixed a Wi-Fi issue in a heterogeneous setup, which turned out to be caused by a stray DHCPv6 server. - Paste C code, along with an error log, and ask to fix the problem. - Past…

> Keep in mind that ChatGPT has been seriously and intentionally downgraded since March

I found it getting worse for a bit but lately it's been really good. I would say it's being constantly updated, not necessarily getting worse.

Re: Ask HN: Is anyone else bearish on OpenAI?

#286
post #56

I'm not bearish on OpenAI or AGI in general but I'm extremely meh about it. I'm not chomping at the bit to use it like so many are, and I constantly feel like a huge luddite or something for not being super excited about it. The value and time it saves makes sense for folks who struggle with a search engine (many) or doing tasks that are typically considered menial, like writing emails or coding boilerplate. However,…

> The value and time it saves makes sense for folks who struggle with a search engine

You don't need to "struggle" with Google to get value out of it, you simply need to value your time.

If you want an answer to a question, why waste time reading through pages of search results when you can have an AI do that for you, reporting what it finds?

No, it's not perfect but it's pretty damn useful.

Re: Ask HN: Is anyone else bearish on OpenAI?

#288
post #36

Earlier quoted context omitted.

Are you able to give a specific example of a problem it helped you solve? Especially one that you were at a complete blocking point, it provided some solution, and then you were able to continue to expand upon that solution. I keep reading responses like yours, but I haven't seen any specific examples of problems being solved, so it all sounds very abstract. In my interactions with ChatGPT, it felt like just interact…

I don’t need it to solve problems where I’m completely stuck for it to be worth $20 a month. Things I’ve gotten value out of in the past week or two: • Writing a job description • Making a python script to automate some stuff in Asana • Simplifying some management concepts so I could slack them to a coworker All of these are things I could easily do myself. But with ChatGPT, they’re done 75% as well in 10% of the tim…

This is the sort of stuff I use it for all the time, great examples. Or summarizing new concepts, reformatting strings or tables quickly, asking what some strange code snippet might be doing.

Re: Ask HN: Is anyone else bearish on OpenAI?

#289
post #36

Earlier quoted context omitted.

Are you able to give a specific example of a problem it helped you solve? Especially one that you were at a complete blocking point, it provided some solution, and then you were able to continue to expand upon that solution. I keep reading responses like yours, but I haven't seen any specific examples of problems being solved, so it all sounds very abstract. In my interactions with ChatGPT, it felt like just interact…

Are you using GPT-3.5 or GPT-4? A hallucination like FableGL sounds like 3.5, which is not something you should base your expectations of the tech on. Also, although I haven't seen it before, the Warren Abstract Machine seems like too big of a job for GPT-4. It excels at smaller tasks such as "Convert this C code to idiomatic Rust", or "Write a SQL query that grabs records from X, joins with Y, filters by Z". You mig…

> A hallucination like FableGL sounds like 3.5, which is not something you should base your expectations of the tech on.

So in order to properly evaluate a service, we need to already be subscribed to it in the first place?

Re: Ask HN: Is anyone else bearish on OpenAI?

#290
post #36

Earlier quoted context omitted.

Are you able to give a specific example of a problem it helped you solve? Especially one that you were at a complete blocking point, it provided some solution, and then you were able to continue to expand upon that solution. I keep reading responses like yours, but I haven't seen any specific examples of problems being solved, so it all sounds very abstract. In my interactions with ChatGPT, it felt like just interact…

Are you using GPT-3.5 or GPT-4? A hallucination like FableGL sounds like 3.5, which is not something you should base your expectations of the tech on. Also, although I haven't seen it before, the Warren Abstract Machine seems like too big of a job for GPT-4. It excels at smaller tasks such as "Convert this C code to idiomatic Rust", or "Write a SQL query that grabs records from X, joins with Y, filters by Z". You mig…

So funny SQL query example, do you realize you could describe your exact query to the engine itself in the more precise and short terms? Writing "select X from Y where Z join XX" seems much more practical and rational than using this whole sentence "Write a SQL query that grabs records from X, joins with Y, filters by Z" to achieve the same.
Post reply on HN