So, what do you do when a human makes a mistake that causes a runaway reaction? Are you telling me you're expected to fight that fire manually, or does your lab have automated fire suppressant? Consider this, if no humans were expected to be in the building, and that happened, don't you think that really good fire suppressant systems could be installed, specifically the kind that would otherwise be fatal to any humans still around? Industrial chemical plants have already solved this problem, why would you not be able to apply the same principles to experimental labs?
There are many things that can go wrong. It takes about 10 years to train a synthetic organic chemist, typically a smart and motivated person, to carry out tasks that involve handling hazardous chemicals and more importantly getting the chemistry right. Can a robot cook a recipe from scratch, to perfection, in the kitchen? If no, then it can't do chemistry.
Yes... That is literally what they're best at. How do you think cooked foods are mass produced? And in fact, chemistry is even simpler than cooking because the starting materials are all uniform and consistent. I respect your opinion as a chemist in that there are things that require human intervention right now, but you really underestimate how far technology has come along, and how quickly they will make manual tasks obsolete. Driving is actually more complicated than chemistry from an AI standpoint due to conditions being essentially random, and we already have AI that's close yo completely autonomous driving.
Yes, but those procedures are quite mechanical, and are a prime target for automation. All of that data would already be available to an AI, forever, so it would quickly be able to develop reactions based on the data it already knows. Even more so if we have even a slightly predictive model of quantum chemistry. Another example. Right now, they are training AI to take in medical test data and make diagnoses. That is actually a much more challenging problem than designing reactions as the data quality is poor, and problem space is much much larger. Conversely, chemistry is basically applied thermodynamics, which can (at a high level) be modeled by the Gibbs equation.
So, what do you do when a human makes a mistake that causes a runaway reaction? Are you telling me you're expected to fight that fire manually, or does your lab have automated fire suppressant? Consider this, if no humans were expected to be in the building, and that happened, don't you think that really good fire suppressant systems could be installed, specifically the kind that would otherwise be fatal to any humans still around? Industrial chemical plants have already solved this problem, why would you not be able to apply the same principles to experimental labs?
There are many things that can go wrong. It takes about 10 years to train a synthetic organic chemist, typically a smart and motivated person, to carry out tasks that involve handling hazardous chemicals and more importantly getting the chemistry right. Can a robot cook a recipe from scratch, to perfection, in the kitchen? If no, then it can't do chemistry.
Yes... That is literally what they're best at. How do you think cooked foods are mass produced? And in fact, chemistry is even simpler than cooking because the starting materials are all uniform and consistent. I respect your opinion as a chemist in that there are things that require human intervention right now, but you really underestimate how far technology has come along, and how quickly they will make manual tasks obsolete. Driving is actually more complicated than chemistry from an AI standpoint due to conditions being essentially random, and we already have AI that's close yo completely autonomous driving.
Yes, but those procedures are quite mechanical, and are a prime target for automation. All of that data would already be available to an AI, forever, so it would quickly be able to develop reactions based on the data it already knows. Even more so if we have even a slightly predictive model of quantum chemistry. Another example. Right now, they are training AI to take in medical test data and make diagnoses. That is actually a much more challenging problem than designing reactions as the data quality is poor, and problem space is much much larger. Conversely, chemistry is basically applied thermodynamics, which can (at a high level) be modeled by the Gibbs equation.