I have high hopes for reasoning as the next leap forward now that pre-trained has tapped out.

While scaling laws appear to face challenges, innovation in techniques and approaches may sustain #Artificial Intelligence advancements. The pursuit of superintelligence continues, requiring vigilance and adaptation to its societal and ethical implications.

https://www.platformer.news/openai-google-scaling-laws-anthropic-ai/?ref=platformer-newsletter

AI companies hit a scaling wall

Artificial Intelligence

I have high hopes for reasoning as the next leap forward now that pre-trained has tapped out.  While scaling laws appear to face challenges, innovation in techniques and approaches may sustain #Artificial Intelligence advancements. The pursuit of superintelligence continues, requiring vigilance and adaptation to its societal and ethical implications.  https://www.platformer.news/openai-google-scaling-laws-anthropic-ai/?ref=platformer-newsletter

Cyril

A deep dive on scaling laws finds that even if the traditional model size-power-compute combination is showing diminishing returns, companies are finding lots of new ways to scale and deploying them all.

https://semianalysis.com/2024/12/11/scaling-laws-o1-pro-architecture-reasoning-training-infrastructure-orion-and-claude-3-5-opus-failures/?ref=platformer.news

A deep dive on scaling laws finds that even if the traditional model size-power-compute combination is showing diminishing returns, companies are finding lots of new ways to scale and deploying them all.  https://semianalysis.com/2024/12/11/scaling-laws-o1-pro-architecture-reasoning-training-infrastructure-orion-and-claude-3-5-opus-failures/?ref=platformer.news

There has been an increasing amount of fear, uncertainty and doubt (FUD) regarding AI Scaling laws. A cavalcade of part-time AI industry prognosticators have latched on to any bearish narrative the…

Scaling Laws – O1 Pro Architecture, Reasoning Training Infrastructure, Orion and Claude 3.5 Opus “Failures”

Todd Smith

“Babies double in size every month or two until they don’t,”
“Most exponentials don’t continue indefinitely.”

“Babies double in size every month or two until they don’t,” “Most exponentials don’t continue indefinitely.”

Nate

Google CEO Sundar Pichai said Wednesday that gains in conversational artificial intelligence capabilities would be harder to come by in the coming years, even as he touted the company’s improvements in high-priority areas such as reasoning. The comment appeared to confirm a November report from The Information and mirrored a slowdown that OpenAI was also experiencing from using traditional techniques to improve AI.

“The low-hanging fruit is gone,” Pichai said of the company’s AI research efforts. “The hill is steeper.”

Pichai said models would continue to get better at completing a sequence of actions, which would help with automating software tasks, but “deeper breakthroughs” would be needed for more dramatic improvements.

Pichai also threw barbs at Microsoft, noting that the rival was “using someone else’s models”—a reference to Microsoft’s use of OpenAI technology in its Copilot products. Microsoft has recently prioritized developing its own models, MAI, and Pichai said he “would love to do a side-by-side comparison of Microsoft’s own models and our models, any day, any time.”

https://www.businessinsider.com/google-ceo-sundar-pichai-ai-wall-progress-harder-2024-12

Google CEO Sundar Pichai said Wednesday that gains in conversational artificial intelligence capabilities would be harder to come by in the coming years, even as he touted the company’s improvements in high-priority areas such as reasoning. The comment appeared to confirm a November report from The Information and mirrored a slowdown that OpenAI was also experiencing from using traditional techniques to improve AI.  “The low-hanging fruit is gone,” Pichai said of the company’s AI research efforts. “The hill is steeper.”  Pichai said models would continue to get better at completing a sequence of actions, which would help with automating software tasks, but “deeper breakthroughs” would be needed for more dramatic improvements.  Pichai also threw barbs at Microsoft, noting that the rival was “using someone else’s models”—a reference to Microsoft’s use of OpenAI technology in its Copilot products. Microsoft has recently prioritized developing its own models, MAI, and Pichai said he “would love to do a side-by-side comparison of Microsoft’s own models and our models, any day, any time.”  https://www.businessinsider.com/google-ceo-sundar-pichai-ai-wall-progress-harder-2024-12

Google CEO Sundar Pichai says that he doesn't believe there is an AI "wall" but says new breakthroughs are needed.

Google CEO Sundar Pichai says AI progress will get harder in 2025 because 'the low-hanging fruit is gone'

Sam Mendoza

Last Thursday, an experimental new version of Google’s Gemini model surprised some by immediately vaulting to the top of a closely watched AI leaderboard, placing it ahead of ChatGPT 4o.

https://x.com/andrewcurran_/status/1857113051063005290?s=46&ref=platformer.news

 #Artificial Intelligence

Last Thursday, an experimental new version of Google’s Gemini model surprised some by immediately vaulting to the top of a closely watched AI leaderboard, placing it ahead of ChatGPT 4o.  https://x.com/andrewcurran_/status/1857113051063005290?s=46&ref=platformer.news   #Artificial Intelligence

Cyril

A group of likeminded people in Artificial Intelligence are talking about this.