Raw LLM Responses

@ChipZo- I mean I think the rise will be a continuous thing like we see with all technology. First we had a phone with a touchscreen, now we have that same touchscreen but nicer software, bigger screens, better battery, etc. AI and robotics are separate but related fields with some crossover. In terms of language models like chatgpt, there is currently stuff that they can't answer very well (for example you can give some engineering design problems that they will fail miserably at). Im not an expert in how these systems need to be improved and where they might be applied, but one constraint is the cost to actually build a model like the one behind chatgpt. Those neural networks are "trained" which basically means a process has to sit on a computer processing a bunch of data, turning it into some mathematical problem, and iteratively trying to solve that problem. That takes a lot of computational resources and also a lot of energy. Only the richest companies can really afford to build these and if a more specific model wants to be built, like lets say there's some logging specific chatgpt, it would be great if that model could be built without costing hundreds of millions. In robotics, there are a ton of issues. I have one example of picking a tomato. If you have a robotic hand trying to pick the tomato, it doesn't have the same feedback of feeling the firmness and knowing not to grip it too hard and break it for one tomato vs another tomato being super firm and being able to grip it hard. There has to be some kind of real time feedback with the haptic sensors to try and allow a robot to be able to handle that decision making in real time. That's just one small example too. How can you make one algorithm that allows a robot to pick tomatoes, oranges, and apples without damaging them? It sounds simple but its not. For walking robots, how can we verify safety when operating in human environments? What if its at a house its never been to before and all the sudden one of its cameras fail? Will it still be able to meet its task? Will it crash into a human and kill them all the sudden because its lost some vision? Characterizing failures and how to be resilient to them is a topic of research for robots. There are tons more like this, hopefully I gave you an idea but if you want to know more im happy to go more into the weeds.