5 ms·
It would be interesting to see how a general-purpose AI like GPT-4 trained on internal corporate knowledgebase would perform. Theoretically, such an AI agent w
by spaceman_2020 4y ago
It would be interesting to see how a general-purpose AI like GPT-4 trained on internal corporate knowledgebase would perform.
Theoretically, such an AI agent would know your internal policies, biases, guidelines. It would also have more knowledge about the company than any individual worker, collating data from across divisions.
In large corps, the silofication of knowledge is a real problem. But an AI agent trained across the company's functions would not have this issue.
- _glass 4y agoEspecially with all of the meetings switching to virtual spaces with transcripts. This would be so powerful. When you can ask (or create documentation) based on all of those meetings. Then enriched by all of the chat messages, the company's code. Best thing, it doesn't even have to run in the cloud. As it looks like it can be an on-premise inference machine. The future is bright for surveillance.
- shaky-carrousel 4y agoIt sure looks like a beautiful future. I'll do my best to make it a reality as soon as possible.
- spaceman_2020 4y agoStretching that further, it is entirely possible to track every single interaction you have with your work computer. You can track your mouse movements, slack messages, work email, keystrokes... With this data, it is possible to create a digital version of you, the employee - a digital version who writes like you, moves his mouse like you, and even makes the same coding errors as you. I can totally see a near future where companies create digital avatars of star employees that stay with the company in perpetuity.
- _glass 4y agoThat sounds like an amazing sci-fi plot, and somehow doesn't seem so far off from the future. Only in Germany where I work I am sure this will be blocked by worker's regulations. But how long can they do that, if you have already synth workers from all around the world. We can start a recruiting agency for synths. The more I work with LLMs the more I am aware of the synthetic potential.
- spaceman_2020 4y agoI can totally see people making digital clones of themselves and sending them out there to work multiple jobs. The future is going to be very...interesting.
- ChatGTP 4y agoI know this is cynical but it’s honestly quite rare anywhere I work had anything truly important written down that’s current. I know that sounds bad but it’s true, at least I my experience. I think the issue here is, how do you keep this model up to date ? Maybe that’s a trivial thing to do ?
- thefounder 4y ago>> how do you keep this model up to date ? You could place mics everywhere. The machine will hear everything and know everything unless you whisper
- nullsense 4y agoSo this is why OpenAI named it Whisper.
- shaky-carrousel 4y agoWell, seeing how speech recognition is improving, I think that the whispering issue is going to be fixed sooner than later.
- bryanrasmussen 4y ago>Theoretically, such an AI agent would know your internal... biases, I think it would follow your internal biases, but would not know them. It is an interesting question as to how to train a model to identify your internal biases.
- phkahler 4y agoWho trains it? How often? All those internal procedures are finally going to have to make sense because they are going to be interpreted by a computer, not half ignored by humans.