New platform increases transparency and reliability of neural networks
Modern artificial intelligence (AI) systems like ChatGPT can provide convincing answers to any question. However, they don’t always admit their mistakes or point out areas of uncertainty in their knowledge. This is a problem, especially when AI is used in important areas, such as drug development, data analysis, or driving autonomous cars.
/imgs/2025/06/04/10/6839386/6051e9c9ce4bfab7afd7369b998d22dbc94f1e30.jpeg)
The new Capsa platform from Themis AI helps solve this problem. It can run any machine learning model and quickly identify questionable or erroneous answers, correcting them before they cause serious problems. Capsa “wrapped” the model and analyzed where it might be wrong or uncertain, helping to improve the accuracy and reliability of AI.
Themis AI was founded in 2021 by scientists from the Massachusetts Institute of Technology (MIT). The company already collaborates with companies from various industries: from telecommunications and oil production to pharmaceuticals. Their technologies help reduce forecast errors and make AI work more transparent and predictable.