The idea of computers and robots greatly accelerating the rate of progress in chemical and drug research is not science fiction.
Autonomous chemical research with large language models
Transformer-based large language models are making significant strides in various fields, such as natural language processing1,2,3,4,5, biology6,7, chemistry8,9,10 and computer programming11,12. Here, we show the development and capabilities of Coscientist, an artificial intelligence system driven by GPT-4 that autonomously designs, plans and performs complex experiments by incorporating large language models empowered by tools such as internet and documentation search, code execution and experimental automation. Coscientist showcases its potential for accelerating research across six diverse tasks, including the successful reaction optimization of palladium-catalysed cross-couplings, while exhibiting advanced capabilities for (semi-)autonomous experimental design and execution. Our findings demonstrate the versatility, efficacy and explainability of artificial intelligence systems like Coscientist in advancing research.
Nature
It seems to me that the speed limit here is not anything imposed by the computers and robots, but your ability to measure progress and give the computers and robots feedback. With chemicals, you could tell the robots to find a combination of compounds that will do XYZ, where XYZ is something you can measure like an amount of energy or a color. With drugs, your issue could be how to test the results to see if they are working. If you test them on a computer model, your ability to measure depends on how good the computer model is. Let’s say you wanted to breed a super-intelligent mouse. There should be ways to measure the intelligence of a mouse. So you could take 100 mice test them all, find the two smartest and create a new batch of 100 embryos from the smartest male and female (or maybe at some point gender is no longer a limitation?). Now you have to wait for those 100 embryos to grow up to the point you can repeat the process. The limiting step here would be how long it takes the mice to develop to the point they can be tested. If they could somehow be tested at the embyro stage, maybe you could create a thousand generations of mouse directed mouse evolution in a matter of hours or days? Well, then, you can let the super-intelligent mice design the next round of robots.