Over that time really one of the consistent conclusions is that more attention should be payed to how different people handle bench science.
There appears to be at least a 10x difference and many times way more when it comes to effectiveness and reliability.
I love math but this should not actually be measured numerically, it should be intuitively gathered in such a way as to put actual measurement to shame. If that doesn't work the first time, then regather, you know what the measurements are and how far they need to be exceeded so don't settle for less.
I would put my money on having one of the 20x-plus operators personally train the lab AI, rather than expect passive uptake of published data to make as much difference.
This has been on my mind for over 45 years.
Over that time really one of the consistent conclusions is that more attention should be payed to how different people handle bench science.
There appears to be at least a 10x difference and many times way more when it comes to effectiveness and reliability.
I love math but this should not actually be measured numerically, it should be intuitively gathered in such a way as to put actual measurement to shame. If that doesn't work the first time, then regather, you know what the measurements are and how far they need to be exceeded so don't settle for less.
I would put my money on having one of the 20x-plus operators personally train the lab AI, rather than expect passive uptake of published data to make as much difference.