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

What we learned in 6 months of working on an AI Developer

blog.pythagora.ai

Post reply on HN
Print view View original

41–50 of 51 posts

Re: What we learned in 6 months of working on an AI Developer

#41

Post by jaggederest »

c0balt wrote: post #38
samus wrote: post #17

Earlier quoted context omitted.

That should be quite easy compared to software development, which is much more open-ended since the requirement are usually more nebulous, potentially contradictory, and at times simply wrong.

> That should be quite easy SysAdmin stuff is quite easy in terms of complexity to some sw stuff. The problems, similar to traditional engineering, tend to come from the rather high cost of failure. To expand further, it's easy to setup a system but hard to setup one that's reliable and/ or resilient. It's hard to maintain systems that are not documented and/ or wrongly documented (outdated, inaccurate). It's even ha…

Yeah imagine giving an AI root access to the server with your production database on it. Now I really want to try that in a VM to see what will happen. Even odds at some point it tries to rm-rf everything.

Re: What we learned in 6 months of working on an AI Developer

#42

Post by bsenftner »

stevage wrote: post #4

The focus on upfront specs feels a bit off. Since it's apparently cheap to generate running code, as a user, I'd much rather be able to just iterate really fast and use output to refine my requirements rather than having to laboriously state them all up front. Agile rather than waterfall if you will.

It's the LLM that needs the upfront specs, regardless of what you'd like, at this point. If that is the case, implement a nondestructive composable system, like node based visual editors - ComfyUI for example. Change the upfront specification "node" and let the LLM cascade through that and any attached nodes creating the code (or whatever) fresh each time.

Re: What we learned in 6 months of working on an AI Developer

#43

Post by threeseed »

throwaway2990 wrote: post #39
threeseed wrote: post #26

Earlier quoted context omitted.

> China is not similarly constrained and so could end up outcompeting the U.S. With what technology ? US has long term export controls on China and as they have demonstrated with Russia recently once you have secondary sanctions in place everyone falls into line. So it's pretty likely they will be effective.

The export controls simply don’t work. Those chips still make their way into China. All those export controls do is slow it down a bit. But China can outfit itself with more hardware even if it’s not as fast as the latest iteration and still speed past the U.S. while the U.S. and the EU argue about AI being racist or not.

> Those chips still make their way into China

Not at the scale you need to build a world-class AI.

Let me know when Huawei is able to place an order for 300,000 GPUs.

Re: What we learned in 6 months of working on an AI Developer

#44

Post by bugglebeetle »

threeseed wrote: post #43
throwaway2990 wrote: post #39

Earlier quoted context omitted.

The export controls simply don’t work. Those chips still make their way into China. All those export controls do is slow it down a bit. But China can outfit itself with more hardware even if it’s not as fast as the latest iteration and still speed past the U.S. while the U.S. and the EU argue about AI being racist or not.

> Those chips still make their way into China Not at the scale you need to build a world-class AI. Let me know when Huawei is able to place an order for 300,000 GPUs.

Okay:

https://asia.nikkei.com/Business/Business-Spotlight/How-Chin...

Re: What we learned in 6 months of working on an AI Developer

#45

Post by layer8 »

withinboredom wrote: post #12
Lerc wrote: post #7

Even though I don't think GPT-4 is up to the task, it does seem like now is the right time to be working on these things. Pretty soon GPT-4 will not be the best in the field. The next generation will perform much better. Possibly the most frustrating thing I find about GPT-4 is how close it gets with it's wrong answers. It's easy to dismiss a lesser answer when it responds with a laughably out-of-band idea. GPT-4 oft…

Oh man. When it’s so close but wrong it’s amazing for creative endeavors! For technical ones, it is quite a bad thing. It’s like being a Star Wars fan but the AI just wants to talk about Star Trek. I think this is why the non-tech people see AI as so amazing. For anything human and non-technical, the “almost but not quite” nature is a good thing. I was using an AI to help me debug a weird thing (mainly summarizing lo…

What’s frustrating is that the one reason I ever wanted AI is to have a Lt. Cdr. Data or ship computer equivalent that is logical and correct to a fault and that helps me reason through things, but what we got now is almost exactly the opposite, we have to help it reason through things and have to double-check everything for correctness.

Re: What we learned in 6 months of working on an AI Developer

#46

Post by fragmede »

amelius wrote: post #6

Until I see an AI sysadmin that can help with basic configure/make problems, I don't have high hopes for an AI developer.

https://chat.openai.com/share/c424f444-c7ac-476f-bd10-02234e...

I picked a random GitHub issue that was some issue with ./configure. Seems like it helps to me.

Re: What we learned in 6 months of working on an AI Developer

#47

Post by samus »

jaggederest wrote: post #41
c0balt wrote: post #38

Earlier quoted context omitted.

> That should be quite easy SysAdmin stuff is quite easy in terms of complexity to some sw stuff. The problems, similar to traditional engineering, tend to come from the rather high cost of failure. To expand further, it's easy to setup a system but hard to setup one that's reliable and/ or resilient. It's hard to maintain systems that are not documented and/ or wrongly documented (outdated, inaccurate). It's even ha…

Yeah imagine giving an AI root access to the server with your production database on it. Now I really want to try that in a VM to see what will happen. Even odds at some point it tries to rm-rf everything.

Nobody is considering giving AI unrestricted access to anything yet. Code written by AI is reviewed by humans, and I would be shocked to hear that sysadmins are considering letting AI agents execute arbitrary commands.

Re: What we learned in 6 months of working on an AI Developer

#48

Post by singularity2001 »

samus wrote: post #47
jaggederest wrote: post #41

Earlier quoted context omitted.

Yeah imagine giving an AI root access to the server with your production database on it. Now I really want to try that in a VM to see what will happen. Even odds at some point it tries to rm-rf everything.

Nobody is considering giving AI unrestricted access to anything yet. Code written by AI is reviewed by humans, and I would be shocked to hear that sysadmins are considering letting AI agents execute arbitrary commands.

I gave GPT root access, so far without regrets

admittedly I recently added a soft barrier "for dangerous commands please ask for confirmation"

Re: What we learned in 6 months of working on an AI Developer

#49

Post by samus »

singularity2001 wrote: post #48
samus wrote: post #47

Earlier quoted context omitted.

Nobody is considering giving AI unrestricted access to anything yet. Code written by AI is reviewed by humans, and I would be shocked to hear that sysadmins are considering letting AI agents execute arbitrary commands.

I gave GPT root access, so far without regrets admittedly I recently added a soft barrier "for dangerous commands please ask for confirmation"

Username checks out!

Re: What we learned in 6 months of working on an AI Developer

#50

Post by singularity2001 »

ctoth wrote: post #13

One of the things they seem to have figured out is the requirement to at least model a sort of actor-critic architecture with their agents. It helps quite a bit. They seem to badmouth Aider a tad (not cool) but I do wonder how a full-stack of this + Aider might work? There needs to also be some sort of good test generator involved. All that said, any time someone actually demonstrates progress on the automated Softwa…

Mentioning Aider, which other tools attempt to act as complete coding copilots? Github Copilot pro seems to have ended some of their experiments, or did I just not get access to the right beta?
Post reply on HN