|
|
|
|
|
|
by Havoc
803 days ago
|
|
|
> the systems don’t have a path to profitable functionality. I don’t buy this at all. Sure loads of ChatGPT wrappers will die but there is definitely money to be made. Robotics is going to be big. Companies are firing customer service people. Etc Sure that last one is perhaps not desirable but it does prove both practical value and feasibility And as the price per token plummets and context increases more application will open up. Just because VC money is keeping some shops open that should die doesn’t mean that’s true for the entire field |
|
|
And they're starting to reap the rewards:
https://www.forbes.com/sites/marisagarcia/2024/02/19/what-ai...
AI customer service chatbots are just as prone to hallucination as any other AI chatbot, and companies are finding out that they're liable when their AIs lie to their customers. As we see more companies replacing humans with AIs, we're going to start to see more companies being found liable for the claims that their AIs spun out of whole cloth.
We haven't seen an AI yet which can consistently reproduce the truth, rather than something that sounds a lot like something that could be the truth. When someone makes an AI which is reliably just a knowledge engine, which can take an input, process it, and accurately convey the knowledge available in that input, then that will be a killer app; when an AI can take a company's documentation and summarize and condense it for easier consumption, that AI will be insanely popular in business cases both internal and external. Until then, these customer-service-replacing AI chatbots are going to prove to be more and more of a maintenance and liability headache.