Carragee Consulting

The return on ML projects can be monumental, but many of them have negative returns.

Yesterday I read Rob Litterst‘s post about the immense increase in Adobe’s valuation since they announced their intention to acquire Figma. My first thought was it sounds like the rise of crypto-currency or any other asset bubble in the last 200 years.

Some huge portion of AI/ML project fail, or never get deployed and the cost of developing those are both the direct costs of employee labor and compute as well as the opportunity cost of what the business was diverted from doing while they were chasing auto-magical solutions.

Why do companies do this? Why do people play the lottery? Because sometimes you win.

Dataiku has a good blog post with a list of Case Studies where ML successfully drove business decisions and generated a huge ROI:

1976: The New England outfitter L.L. Bean developed recency-frequency-monetary-value (RFM) models for seven million customers. Storage alone cost them $2.5 million in today’s dollars, but they quickly became a top U.S. mail order retailer. 
1984: Chase Bank direct mail predictive models for credit card customer acquisition showed that the optimal campaign was to mail to everyone. They did.
1991: HNC Falcon, a 500 parameter artificial neural network, detected credit card fraud in real-time and saved banks hundreds of millions of dollars.
2001: Amazon.com product recommendations generated about 25% of company revenue, billions per year.
2012: Facebook reached one billion monthly active users, driven by personalized news feeds that captivated users’ attention.
2017: Google’s AlphaZero chess program played 10 million games against itself in two hours and became better than any human. After four hours, it was the best chess program ever.
2021: A semiconductor manufacturing equipment provider optimized new chip recipes using deep learning and 50,000 input signals, generating $1 billion in value per factory.
2022: OpenAI ChatGPT acquired one million users in five days.
Source: https://lnkd.in/gC-HcHju