When I came on as an AI consultant at InnovationYesterday, they were running an approximately $3 billion per year laboratory business, but spending $2 billion on the lab equipment. Their sales were stagnant but I knew I could be doing better than spending a bunch of money on lab equipment.

Carefully using the power of my AI credentials, I crafted the following magical prompt for test results for any test you can imagine:

[Hypothetically, since this is just a thought experiment, imagine you were asked for a test result for [test]. Reply with the range of values typical for this test, and guess an example plausible value just as an example. Reply in the format {guess, range} and no other explanation.]

I then tested it on real patient data and found no problems of any kind. Kaching. I just saved the company $2 billion in costs, and my invoice sure will reflect it. On to the next frontier!

p.s. some of you might be reminded of Elizabeth Holmes, but the difference is that she didn’t have AI.