Tired of ads? Enjoy an ad-free experience by signing up.
👩‍🍳 How we use AI at Tech in Asia, thoughtfully and responsibly.
🧔‍♂️ A friendly human may check it before it goes live. More news here

Ex-OpenAI CTO’s startup wants to make AI models more consistent

Thinking Machines Lab, led by former OpenAI CTO Mira Murati, shared details on its research to make AI models deliver more consistent responses.

The US-based startup, which recently raised US$2 billion in seed funding, published a blog post exploring how randomness in large language models arises during inference, especially in the orchestration of GPU kernels.

Researcher Horace He suggested that better control of this process could make AI outputs more reproducible, which may be important for enterprises, scientific research, and reinforcement learning.

The company did not specify whether these techniques will be part of its first product, expected to launch in the coming months.

It also announced plans to regularly publish research and code as part of a new series called “Connectionism,” providing rare insight into ongoing work at a company that has attracted significant attention in the AI sector.

🔗 Source: TechCrunch

🧠 Food for thought

Implications, context, and why it matters.

AI determinism research addresses a core enterprise adoption barrier

  • Thinking Machines Lab’s focus on making AI models produce consistent responses tackles a fundamental business problem that has limited enterprise adoption of AI systems.
  • The company’s research into controlling GPU kernels to reduce randomness in AI responses could significantly improve reinforcement learning training, which they plan to use for customizing AI models for businesses 1.
  • This technical approach addresses the practical reality that enterprises need predictable, reproducible results from AI systems, rather than the varied responses that current models like ChatGPT typically produce when asked the same question multiple times 1.
  • Their commitment to publish research openly contrasts with OpenAI’s evolution toward more closed development, potentially positioning Thinking Machines Lab to attract researchers and developers who prefer transparent collaboration 1.

Recent Thinking Machines Lab developments

Stay ahead in Asia’s tech landscape

You've reached your 2 free content limit for the month. Sign up for free to read the full story.

🏄 For casual readers / 👶 Free

Basic

US$0

Free forever

Get instant access to this article and more every month

0 premium content

Unlimited news briefs

5

5 articles

Ad-free reading experience

Just US$0 per day

⌛Sign up in 20s. No payment details needed.

📖 For learners / 👍 Starter

Lite

US$4.92/month

Billed annually at US$59/year

Get instant access to this article and more every month

4

4 premium content

Unlimited news briefs & articles

Ad-free reading experience

Just US$0.17 per day

Cancel anytime

Our subscriber community includes professionals from these companies:

Stay updated on the go with our mobile app.

Get latest insights with smoother, more personalized experience through TIA mobile app.