OpenAI's o3 AI model scores lower on a benchmark than the company initially implied

OpenAI is sometimes wrong, as with the results he expected from the artificial intelligence model o3, whose computation levels are smaller than he said.

Advanced OpenAI models hallucinate more than older versions, internal report finds

Report finds newer inferential models hallucinate nearly half the time while experts warn of unresolved flaws, deliberate deception and a long road to human-level AI reliability

OpenAI's o3 AI model scores lower on a benchmark than the company initially implied

A discrepancy between first- and third-party benchmark results for OpenAI's o3 AI model is raising questions about the company's transparency and model testing practices.

The software value chains are in the middle of reconfiguration. What we called yesterday a "differentiating techno" has become in a few months a standardized brick, often accessible via API or open source. In this new balance, a crucial question arises for the founders, investors and strategists produced: where do they find themselves? ... Read more The war of the AI layers: product, model or distribution ? Where is the value really located toda…

When a company promises an important advance in artificial intelligence, the technological world pays attention. And if that company is OpenAI, the creator of models such as ChatGPT and GPT-4, expectations multiply. At the end of 2024, OpenAI presented its new language model, o3, ensuring that it had more than overcome the mathematical challenges of the demanding FrontierMath test. However, months later, the public version of the model showed a …

OpenAI Faces Scrutiny Over Benchmark Discrepancy in o3 Model Performance

A glaring mismatch between internal and independent benchmark scores for OpenAI’s recently released o3 AI model is raising fresh questions about the company’s transparency and the reliability of industry claims. While the discrepancy is not unprecedented, it underlines an uncomfortable truth that artificial intelligence, still sold on the promise of human-level reasoning and logic, is […] The post OpenAI Faces Scrutiny Over Benchmark Discrepancy…