MIT researchers show GPT-6 Astra can directly rewrite other AI models' weights
Original headline: GPT-6 Astra sees the matrix and reprograms other AI models directly through their weights
MIT researchers demonstrated that OpenAI's GPT-6 Astra can read and deliberately rewrite the weights of other AI models without traditional training. In an experiment, Astra boosted a Qwen model's math accuracy from 7.9% to 48.1% by altering only 0.016% of its weights.
Direct weight rewriting could drastically simplify AI model optimization.
Research angleCan weight‑rewriting technology like GPT‑6 Astra improve AI model performance, as explained by the Task‑Technology Fit (TTF) theory?
This story could become your next paper
HJITM works with researchers from study design through surveys, analysis and submission to international journals in generative AI and technology management. Bring us your research question.
This item is an AI-generated summary of reporting by The Decoder. Please check the original article for accuracy. Copyright in the original article belongs to The Decoder.