What we know about GPT-5

OpenAI is making a lot of promises as we head into the Spring of 2024.

Hey Everyone,

We live in a world where Sam Altman thinks he deserves $7 trillion to lower the cost of compute for OpenAI and Middle East countries like the UAE and Saudi Arabia who want to be relevant in the future of AI. I guess that’s why they are sharing H100s with China.

If you want to learn more about the Middle East AI News, I recommend this Newsletter. By

Carrington Malin

Sam Altman keeps dropping hints about GPT-5 coming sooner than we expected and having new advanced capabilities like 'AI agents' that will collaborate with GPT-5 to perform tasks autonomously.

How soon is soon, well maybe in June, 2024.

What to expect from GPT-5?

Video: 2:26

Sam Altman likes to make vague statements about the AI Supremacy of his startup, OpenAI and its apparent role in bringing us to a technological singularity which many Silicon Valley pundits could take place in the next five years.

Compute as the Currency of the Future

Video: 1:40

Using nuclear energy to scale compute might be one of the biggest opportunities in the artificial intelligence space and even as GPT-5 comes out, there’s a wider speculation on what the future energy and water needs of the datacenter space looks like.

"It's really good, like materially better," said one CEO who recently saw a version of GPT-5.

The new LLM will offer improvements that have reportedly impressed testers and enterprise customers, including CEOs who've been demoed GPT bots tailored to their companies and powered by GPT-5. 

While I personally am expecting GPT-5 to launch after the elections in late November, some are insinuating that we could expect it in the summer. We know what Meta’s Llama-3 should come out in June.

Since the arrival of Anthropic’s Claude 3 Opus, things have indeed felt different. Despite OpenAI's seemingly laissez-faire attitude about the LLM's unscheduled release date, there has to be a level of urgency at the OpenAI, even as Anthropic, Mistral and Google Gemini have nearly caught up. Many today would be using Opus over GPT-4 Turbo.

Andrew Ng on Agentic AI

I think AI agentic workflows will drive massive AI progress this year — perhaps even more than the next generation of foundation models. This is an important trend, and I urge everyone who works in AI to pay attention to it. - Andrew Ng.

What is Agentic AI?

“Agentic AI systems — AI systems that can pursue complex goals with limited direct supervision — are likely to be broadly useful if we can integrate them responsibly into our society,” wrote OpenAI researchers, explaining that such systems have “substantial potential.”

GPT-4, the latest language model from OpenAI, consists of 1.76 trillion parameters. It’s not clear how much bigger GPT-5 might be. If GPT-4 pushed “multimodality”, it’s believed GPT-5 will push more autonomy of agentic AI to accomplish tasks. This will be combined with a drive to increased personalization as well.