Elon Musk's new X marketing campaign was a risk th...
news-extra-space
It's a valid concern given that Meta is a major social media platform that has previously been under media and governmental scrutiny for distributing false information. Considering how simple it is to persuade most LLMs to ignore its safety procedures and social filters, unleashing Meta's LLMs too soon might be fatal for the business. In the meantime, Microsoft expanded its collaboration with OpenAI, in which it has made a significant investment since 2019. Microsoft intends to invest billions in the initiative and employ OpenAI's ChatGPT, Dall-E, and Codex AI technology to improve its products. Microsoft's Azure cloud computing network is used only by OpenAI. [caption id="" align="aligncenter" width="640"] Microsoft expands its pact with OpenAI in 'multibillion dollar' deal | Engadget[/caption] Photo Credit: Engadget Although LLMs and many other AI technologies utilized by ChatGPT were not created by OpenAI, it surely seems inventive to have made this revolutionary service accessible in such an unfettered manner before Meta and Google were even considering it. AI is used by Meta for advanced research and to identify false information in its social media networks, and Google has long integrated AI into Android and Google search. That makes all the difference to the public because neither firm has made the capabilities of their AI systems accessible to everyone the way that OpenAI's ChatGPT and Dall-E do. [caption id="" align="aligncenter" width="1200"] OpenAI Captures the Magic of AI With ChatGPT and DALL-E | Built In[/caption] Photo Credit: Built-in The public's impression of ChatGPT is generally favorable, but given that both Meta and Google have hinted that more AI capabilities are on the way, that may soon change.LLMs are still making sh*t up. That's fine if you use them as writing assistants. Not good as question answerers, search engines, etc. RLHF merely mitigates the most frequent mistakes without actually fixing the problem. https://t.co/XnDxF8Q9Zr
— Yann LeCun (@ylecun) January 22, 2023
Leave a Reply