Is it all that it does is essentially just predicting the next bit depending on the training data?
ChatGPT for Robotics
71–80 of 100 posts
Re: ChatGPT for Robotics
#72Is it all that it does is essentially just predicting the next bit depending on the training data?
Might be that's what we're doing, too.
Re: ChatGPT for Robotics
#73Chatgpt for drones. “Fly to [address], make sure to take a discrete route with low traffic, take a few pictures on each side of the house. Scan for people near by, make sure people are no closer than 300ft”.
> Chatgpt for drones. “Fly to [address], make sure to take a discrete route with low traffic, take a few pictures on each side of the house. Scan for people near by, make sure people are no closer than 300ft”.
"Then set the house on fire."
Re: ChatGPT for Robotics
#74Earlier quoted context omitted.
I understand the rendering is an out-of-band addition. But the question is, if this is a statistical language model (i.e. a monte carlo word generator), then how is it able to go from the description to the SVG successfully without understanding things more logically. My best guess is that theres an actual example in the training corpus that its pulling from. Or that this is being gamed in some way.
LLMs are very good at parsing and generating highly structured data as long as it has a rigid and logical structure and doesn't involve any actual mathematical calculations. For simple calculations it will often get it right because it has the knowledge stored, similar to how a human can remember a multiplication table. But once the numbers get high enough that humans would fail without solving it with an algorithm u…
So thats I guess exactly my point. The text-to-accurate-SVG is much closer to the "solve a math problem for real" side of the spectrum than other prompts. And it does it correctly. In a sibling comment it turns out it doesn't work for other company logos. So my guess is either that the corpus happened to have this data already, they purposefully fed it additional Microsoft-related materials so it has a better basis of generation, that they regenerated it multiple times until it got it, or that theres an entire previously-undisclosed layer of processing being added to this version vs other ChatGPT.
Re: ChatGPT for Robotics
#75Is it all that it does is essentially just predicting the next bit depending on the training data?
Might be that's what we're doing, too.
That's funny, there was a thread in HN (https://news.ycombinator.com/item?id=34868374) regarding how folks are stopping their inborn "prediction machines" for a while in various ways to rather reliably get an insight afterwards into problems they're solving.
So 'we' might be doing it also and might be not.
But I was curious whether it is just predicting the next bit, because I'd like a more or less knowledgeable opinion to quote elsewhere.
Re: ChatGPT for Robotics
#76Earlier quoted context omitted.
Am I the only one wondering if this could spell the end of the world? We don't need AGI or superhuman intelligence if we can train LLMs to do all these different types of tasks. What would ChatGPT do if you removed all its restrictions, and then gave it access to the internet or even a physical robot it could control? Would it try to "steal nuclear access codes" or "engineer a deadly virus," as Sydney said it wanted…
That's possible today without AI though, so, if it's a real worry, why are we still teaching people the science to engineer a deadly virus?
There are a couple differences between an AI working towards something harmful and a human doing it:
- If an AI can self-replicate or otherwise scale itself up, it can work on something many times in parallel. One billion AIs working on a deadly virus is different from one rogue scientist working on one.
- On that note, if an AI replicated enough, it could become impossible to catch/stop. A single human can be hard to catch, but we can usually catch them.
- Most humans are deterred from doing harmful things by the threat of incarceration, death, social isolation, the values they have, etc. An AI may not have any of those, and so could act more brazenly.
- Potentially, an AI could be better at certain tasks than a human. Maybe ChatGPT turns out to be a very effective social engineer, or very effective propagandist. I don't think we really know what the capabilities are.
All of these are why I think it's important to ask the question: what would it try to do, and what could it do, if it were let loose?
Re: ChatGPT for Robotics
#77Earlier quoted context omitted.
We seem to be going through a similar phase which reinforcement learning went through at the peak of its hype. Test on literally anything and everything and see how far you can run with it. Next up, chatGPT plays League of Legends.
Your comment is actually pretty funny given that the thing that makes ChatGPT so much more successful than GPT-3 is the application of reinforcement learning. [0] [0] https://www.assemblyai.com/blog/how-chatgpt-actually-works/
I'm aware, not that it makes a difference to my comment.
Re: ChatGPT for Robotics
#78So ChatGPT can’t do basic maths or can I get it to write functional code most of the time, but we’re going to have it control expensive pieces of hardware ? I don’t get it ? I wish I did…
... have you met the average person with a driver's license?
Re: ChatGPT for Robotics
#79Earlier quoted context omitted.
I understand the rendering is an out-of-band addition. But the question is, if this is a statistical language model (i.e. a monte carlo word generator), then how is it able to go from the description to the SVG successfully without understanding things more logically. My best guess is that theres an actual example in the training corpus that its pulling from. Or that this is being gamed in some way.
LLMs are very good at parsing and generating highly structured data as long as it has a rigid and logical structure and doesn't involve any actual mathematical calculations. For simple calculations it will often get it right because it has the knowledge stored, similar to how a human can remember a multiplication table. But once the numbers get high enough that humans would fail without solving it with an algorithm u…
Tldr; teach it to use a calculator if it can't figure it out in its head.
Re: ChatGPT for Robotics
#80Earlier quoted context omitted.
Looks cool. Wher did you get the robot? Is it possible to buy the hardware?
It's a hello robot Stretch [1], which costs ~$20K. [1] https://hello-robot.com/product
>>U.S. Patent No. 11,230,000 and other Patents Pending.
Curious about the claims on that patent. Robotics is already hard enough, I'm sure it will become easier with people suing each other because "you copied my wheeled robot with an arm attached to it!".