Source URL: https://simonwillison.net/2025/Mar/2/ethan-mollick/
Source: Simon Willison’s Weblog
Title: Quoting Ethan Mollick
Feedly Summary: After publishing this piece, I was contacted by Anthropic who told me that Sonnet 3.7 would not be considered a 10^26 FLOP model and cost a few tens of millions of dollars to train, though future models will be much bigger.
— Ethan Mollick
Tags: ethan-mollick, anthropic, claude, generative-ai, ai, llms
AI Summary and Description: Yes
Summary: The communication from Anthropic regarding the Sonnet 3.7 model highlights the cost considerations and performance expectations in generative AI models. This insight is relevant for professionals in AI and cloud computing sectors, particularly those involved in model development and resource allocation.
Detailed Description: The text provides critical insight into the development and financial implications of generative AI models. The interaction with Anthropic contains several noteworthy points:
– **Model Performance and Cost**: Anthropic clarified that Sonnet 3.7 is not a supercomputing-level model (10^26 FLOP) and indicated that its training cost is in the range of tens of millions of dollars.
– **Future Projections**: The mention that future models will be larger implies significant growth in AI model capabilities and associated training costs.
– **Industry Context**: This kind of dialogue reflects the ongoing evolution in the generative AI space, influencing project planning and investment in AI-related initiatives.
Implications for security and compliance professionals are significant:
– **Budgeting for AI Initiatives**: Organizations looking to implement or enhance their AI capabilities need to plan budgets for model training that reflects these insights.
– **Scalability Considerations**: As models grow in complexity and size, infrastructure and security measures will need to adapt to handle increased computational demands.
– **Strategic Planning in AI Development**: Understanding the expected growth and costs can guide decisions on resource allocation and strategic partnerships in the AI landscape.
Overall, the commentary indicates a nuanced understanding of both the technological and financial aspects of AI model development, valuable for stakeholders in AI and cloud computing.