Using ChatGPT + DALL·E 3 to create fake New Yorker cartoons is going to become a regular feature in the email newsletter. Too fun. You can sign up here: https://
aibreakfast.beehiiv.com
@aibreakfast
-
ChatGPT et DALL·E 3 inventent des fausses BD du New Yorker
By
–
-
ChatGPT progresses, Gemini remains behind
By
–
ChatGPT couldn’t solve this a month ago, now it can. Gemini (Bard) still can’t.
-
IA pour traduire vidéos et doubler votre audience
By
–
With AI you can now record video speaking natively in any language, and it even syncs your lips.
— AI Breakfast (@AiBreakfast) 8 décembre 2023
If you are a YouTuber, this will 10x your reach.
Demo featuring @dr_cintas
pic.twitter.com/XDIVzMe55fWith AI you can now record video speaking natively in any language, and it even syncs your lips. If you are a YouTuber, this will 10x your reach. Demo featuring @dr_cintas
-
AI Image Enhancement by Magnific AI
By
–
AI image upscaling from @Magnific_AI turns Midjourney portraits into near perfect photo-quality images.
— AI Breakfast (@AiBreakfast) 8 décembre 2023
Expect your AI videos to be this good by the end of 2024. pic.twitter.com/q1BlnxACdZAI image upscaling from @Magnific_AI turns Midjourney portraits into near perfect photo-quality images. Expect your AI videos to be this good by the end of 2024.
-
Why an AI image of unhappy people in a sauna?
By
–
Then why did they use an AI generated image of unhappy people in a sauna
-
Chain-of-thought: a necessary improvement for LLMs
By
–
Chain-of-thought prompting gives superior responses in LLMs. OpenAI should have implemented that a while ago. That's not "cheating" – that's programming.
-
Google publishes its report on Gemini Ultra
By
–
Google's whitepaper on Gemini Ultra https://
storage.googleapis.com/deepmind-media
/gemini/gemini_1_report.pdf
… -
Gemini Ultra surpasses GPT-4V on benchmarks
By
–
Apparently Gemini Ultra outperforms GPT-4V on almost all benchmark tests. Beats every existing model at coding, and the first to perform better than a human on MMLU (Massive Multitask Language Understanding). Multimodal capabilities support audio and video (never seen before
-
Meta soutient le mouvement open-source IA
By
–
Kinda surprising that Meta is the one backing the open-source AI movement. Glad to see it happening.
-
Meta wants to dominate physical skills learning
By
–
Meta's Ego-Exo4D can make skilled physical tasks teachable through augmented reality
— AI Breakfast (@AiBreakfast) 1 décembre 2023
(e.g. someone wearing smart glasses could be coached through new skills using AI)
Meta will be the future marketplace leader for teaching physical skills, leveraging their $2 billion… pic.twitter.com/U9U2GYpmXOMeta's Ego-Exo4D can make skilled physical tasks teachable through augmented reality (e.g. someone wearing smart glasses could be coached through new skills using AI) Meta will be the future marketplace leader for teaching physical skills, leveraging their $2 billion