Over the weekend, I came across an article in which the president of Google, who has no medical or biology training, claimed that AI will cure cancer in our lifetime. Leaving aside the silliness of describing any complex machine learning algorithm as ‘AI’, what is so frustrating is so many mRNA-based* phase II trials for both cancer and infectious diseases could be funded with a fraction of the money spent on spicy machine learning.
For context, a typical phase II trial costs around $15 million. Leaving the billions of dollars lit on fire for ‘AI’, remember when Facebook was spending $50 million per day to build legless avatars? (they apparently now have legs!).
I would argue the two greatest technological advances** in the last decade are the constant engineering improvements leading to ever cheaper renewable energy and the development of mRNA vaccines, the latter which so far have saved the lives of at least one million Americans, probably more. Instead, we’re burning many, many billions of dollars of capital on a giant mediocrity machine, which seems to be creating more problems than it solves (or ‘solves’).
I will now go outside and yell at clouds.
*Regarding infectious disease, investigating nasal delivery and adjuvants, along with new targets (and various combination thereof), also seems very promising. Meanwhile there are some promising leads for diseases other than cancer too.
**One problem with the term ‘tech’ is that it typically limits the entire discussion to ‘writing code’–which is important–and ignores any kind of physical technology.
mikethemadbiologist
Anyone who claims they’ll “cure cancer” is 110% full of shit since it’s leaking out their noise hole.
People of substance will say something like “We can improve survival rates of non-small-cell carcinoma by 15% if it’s caught early”.
Derek Lowe / In The Pipeline occasionally covers AI/ML in pharma from a practical and realistic perspective, worth reading. For example: https://www.science.org/content/blog-post/latest-automated-analog-generation
Techbros have been promising a digital cure for what ails us since the 80s, coming any day now, make sure to get in on the ground floor with your investments.
LLMs have their roles. Sorting through incomprehensibly vast piles of data to look for patterns? Definitely. Organizing or managing anything at all, let alone a complex effort with many more unknowns than knowns while human lives are on the line? No, never. That should be a criminal offense.
At this point we have the GPU maker paying the model vendor to pay the datacenter operator to pay the GPU maker in an incestuous spin cycle that harvests cash for the people involved while actively destroying public and private goods with a nine-digit butcher’s bill coming due…
“I would argue the two greatest technological advances** in the last decade are the constant engineering improvements leading to ever cheaper renewable energy and the development of mRNA vaccines, the latter which so far have saved the lives of at least one million Americans, probably more.”
I would consider including checkpoint inhibitors, but they are slightly older than a decade.