Nextbigfuture at substack has all of the big AI news for the week.
There was massive information this week in AI, AI startups, AI analysis and AI funding.
1. OpenAI’s Japan CEO reveals GPT-Subsequent shall be launched this 12 months, and its efficient computational load is 100x higher than GPT-4
2. Two $125 Billion AI Coaching Facilities, Beginning at 0.5-1 GW and Scaling to 5-10GW
3. XAI 100k H100 AI knowledge Middle
4. AI Analysis
5. Main current AI funding bulletins together with Protected Superintelligence
1. OpenAI GPT Subsequent and Orion in 2025
GPT-Subsequent shall be launched this 12 months, and its efficient computational load is 100x higher than GPT-4.
GPT-4 NEXT, which shall be launched this 12 months, is anticipated to be educated utilizing a miniature model of Strawberry (for higher reasoning) with roughly the identical computational sources as GPT-4, with an efficient computational load 100 instances higher.
The AI mannequin referred to as ‘GPT Subsequent’ that shall be launched sooner or later will evolve practically 100 instances based mostly on previous efficiency. Not like conventional software program, AI know-how grows exponentially.
This 100 instances enhance doesn’t check with the scaling of computing sources, however relatively to the efficient computational quantity + 2 OOMs, together with enhancements to the structure and studying effectivity.
Orion, which has been within the highlight not too long ago, was educated for a number of months on the equal of 10k H100 in comparison with GPT-4, including 10 instances the computational useful resource scale, making it +3 OOMs, and is anticipated to be launched someday subsequent 12 months.
Two $125 billion AI Coaching Information Facilities
Two firms need to develop synthetic intelligence (AI) knowledge facilities in North Dakota. Commissioner of Commerce Josh Teigen revealed throughout a Public Service Fee assembly in August that two firms had approached him and state Governor Doug Burgum about creating AI knowledge facilities.
The info heart initiatives would begin between 500MW and 1GW initiatives, however might scale as much as 5-10GW services ultimately. The initiatives would price as much as $125 billion every. This could allow the ten,000 higher compute in comparison with what was used for GPT 4. This could allow GPT 6 class LLM (Massive language fashions).
See the substack article for extra.
Brian Wang is a Futurist Thought Chief and a well-liked Science blogger with 1 million readers per thirty days. His weblog Nextbigfuture.com is ranked #1 Science Information Weblog. It covers many disruptive know-how and tendencies together with Area, Robotics, Synthetic Intelligence, Medication, Anti-aging Biotechnology, and Nanotechnology.
Recognized for figuring out innovative applied sciences, he’s presently a Co-Founding father of a startup and fundraiser for prime potential early-stage firms. He’s the Head of Analysis for Allocations for deep know-how investments and an Angel Investor at Area Angels.
A frequent speaker at companies, he has been a TEDx speaker, a Singularity College speaker and visitor at quite a few interviews for radio and podcasts. He’s open to public talking and advising engagements.
