| Personal Webpage | https://yangfengji.net |
| Medium | https://medium.com/@yangfengji |
| My Group @ UVa | https://uvanlp.org |
Nice work from @mattontech! đź‘Ť
There is still a long way to go for using AI in real-world applications
https://spectrum.ieee.org/ai-taking-over-jobs
I agree that it's urgent to have some AI policy in place.
But until then, it does not make much sense to assume the current tools are fine, and only the future systems are dangerous
"Therefore, we call on all AI labs to immediately pause for at least 6 months the training of AI systems more powerful than GPT-4."
There is no reason to believe the danger could ONLY come from a model more powerful than GPT-4.
If the concern is really about "machines flood our information channels", we should also ask OpenAI to pause their APIs for six months.
https://futureoflife.org/open-letter/pause-giant-ai-experiments/
What is the difference between ChatGPT and Bard? Let's hear from themselves.
Highlights:
1. According to ChatGPT, both of them are developed by OpenAI
2. Bard claims it can do a better job of generating text related to the real world.
Here's a glitch/bug/MASSIVE-FSCKING-PRIVACY-RISK associated with the new #Bing #chatbot feature that I haven't seen documented.
Answers from one query bleed over into another.
Earlier today I asked it about the black power salute that Tommie Smith and John Carlos gave at the 1968 Olympic Games.
Tonight I asked it about headlines referencing chatbot halluncinations.
Look what it did. It returned an utterly irrelevant answer, reflecting my previous query. How can this be production technology?
Carl T. Bergstrom