2025 has already brought us the most performant AI ever: What can we do with these supercharged capabilities (and what’s next)?
1 min read
Summary
The Wharton Business School’s Ethan Mollick noted the impressive advances made by large language models (LLMs) recently, including OpenAI’s ChatGPT and Anthropic’s Claude 348037, and xAI’s Grok 3, which have been trained with far more computing power than GPT-4.
Mollick points to two trends as the reason for these advances: the increase in training LLMs and AI’s improving ability to handle complex problems.
These advances could supercharge scientific discovery, cutting research timelines and enabling cross-disciplinary opportunities.
The question is no longer whether AI will transform science, but how quickly its full impact can be realised, and its growing capabilities in reasoning and decision-making present extraordinary promise and formidable challenges.
To navigate these challenges, society must invest in AI governance, education and workforce adaptation.
Mollick concludes by saying if these challenges can be navigated with foresight and responsibility, people and AI can achieve breakthroughs that once seemed impossible.