AH, a VC firm did cause a lot of flutter about the “AI firms” and software as a service via their blogpost. I stumbled on an excellent summary (and a critique below).
My TLDR summary:
- Deep learning costs a lot in compute, for marginal payoffs
- Machine learning startups generally have no moat or meaningful special sauce
- Machine learning startups are mostly services businesses, not software businesses
- Machine learning will be most productive inside large organizations that have data and process inefficiencies
Here’s another fascinating insight:
The promise of “AI” has always been to replace human labor and increase human power over nature. People who actually think ML is “AI” think the machine will just teach itself somehow; no humans needed. Yet, that’s not the financial or physical reality. The reality is, there are interesting models which can be applied to business problems by armies of well trained DBAs, data engineers, statisticians and technicians. These sorts of things are often best grown inside a large existing company to increase productivity.
That brings me to the snake-oil marketing about “personalised medicine”- whose promise has never been realised. It would require immense computational resources to align the EMR data, for instance. The SEER database only reflects entries from the ICD coding (not really patient-reported outcomes) that makes it challenging to apply in clinical practices.
We are still far away from realising the true symbiosis of genetics and clinical outcomes (ideally onboard adaptive radiation therapy) to push the advantage of newer techniques. Currently, modulation and its various types are being pushed out for claiming higher insurance reimbursements that distort the true value obtained by the patients. AI isn’t going to fix any of the structural problems.