AI companies are betting on scientific breakthroughs to justify their existence, but the article questions whether AI is actually delivering on that promise yet

AI companies like OpenAI, DeepMind, and others regularly claim that their large language models (AI systems trained on vast amounts of text) will eventually cure cancer, solve climate change, and make major scientific discoveries—framing these future possibilities as justification for the enormous computing costs and environmental impact of training these systems today.

In practice, LLMs can already help scientists in concrete ways: finding relevant research papers faster, explaining complex concepts, and drafting grant proposals. However, the article's framing suggests these current applications are narrow and incremental, not the transformative breakthroughs the companies promise.

For researchers and industry professionals, this matters because it affects how you should evaluate AI vendors' claims. If a company's survival depends on delivering a scientific moonshot (like curing cancer), their current tools may be oversold relative to their actual capability. You're more likely to get real value from AI for research support tasks—literature review, writing assistance, data interpretation—than from expecting it to independently solve your hardest problems.