Does AI ignore or embrace outliers?
The difference could mean everything!
Penicillin was discovered when Alexander Fleming returned from vacation to find yeasts and molds had contaminated petri dishes in which he had been growing bacteria. But one dish was different: it had a small clear area which led to penicillin. That clear area was an anomaly, an outlier.
Viagra wasn’t originally intended for erectile dysfunction, but scientists noted a side effect that transformed the drug into the blockbuster it is today. Another outlier.
Teflon was similarly produced in an experiment gone sideways.
The microwave. Quinine. X-rays. Velcro. Saccharin. LSD. Play-doh. Vulcanized Rubber. And many other discoveries and inventions were the result of exploring outliers and pursuing mistakes.
When building your AI models, are you eliminating the possibility of examining intriguing outliers?
No comments:
Post a Comment