Mind Matters Natural and Artificial Intelligence News and Analysis
machine-learning-artificial-intelligence-ai-deep-learning-blockchain-neural-network-concept-stockpack-adobe-stock
Machine learning , artificial intelligence, ai, deep learning blockchain neural network concept.

The Man Behind the First Billion-Dollar AI Business

Robert Hecht-Nielsen (1947‒2019) was a significant figure in the second wave of AI. His company specialized in fraud detection
Share
Facebook
Twitter
LinkedIn
Flipboard
Print
Email

OpenAI, the company that fathered ChatGPT, is a multibillion dollar company. As the AI bubble keeps expanding, the news is all over the net.

When using such accolades, news organizations don’t reference the history of previous AI entrepreneurs. OpenAI is not the first billion-dollar AI company.  That honor belongs to one founded almost 40 years ago by Robert Hecht-Nielsen, Ph.D.

HNC software

Robert Hecht-Nielsen (1947‒2019) was a significant figure in the second wave of AI. In 1986 he founded HNC (Hecht-Nielsen Corporation) Software, a company that applied neural networks to problems such as credit card fraud detection. When your credit card was scanned at the local 7-11 in the late 20th century, chances are it was being checked for fraud by HNC software at the other end of the phone line.

HNC Software became highly successful and was acquired by Fair Isaac Corporation in 2002 for $810 million. Adjusted for inflation, that’s $1.4 billion today. If you haven’t heard of Fair Isaac, it’s not a nickname for an Old Testament patriarch.  It’s the “FI” in your FICO creditworthiness score.  

He started with scholarship

Before becoming an entrepreneur, Robert was a scholar. He held an adjunct professorship at the University of California, San Diego. His talks at conferences were inventive and always well attended.  One that I remember was on neural networks.

Neural networks are the backbone of all today’s powerful AI from deep fakes to transformer models like ChatGPT. Neurons in a neural network are connected by weights that let the neurons know how strongly one neuron should affect another.

Training a neural network involves finding the best set of weights to perform a specified task. Robert pointed out that two neural networks with identical performance could have different weights among the neurons. The same neural network, iteratively trained using different starting points, will probably not result in the same final weights even though the final performance is the same. Robert also calculated the number of different ways weights could be arranged and still result in exactly the same performance. The number was combinatorically huge! 

Here’s the bottom line, which may prove useful in litigation:  If my neural network has the exact same weights as your neural network, chances are I copied it from you.

His legacy

Robert co-organized the first major conference on neural networks in 1987. The International Joint Conference on Neural Networks continues to this day.

Robert Hecht-Nielsen singing “Surfin’ USA.”/Robert J. Marks

My most memorable recollection of Robert was a karaoke experience at a 2000 AI conference in Yokohama.  It was in a private room where we were waited on by hostesses dressed in traditional Japanese kimonos. Joining Robert and me were our Japanese friends, two guys from Russia, a German, and a Norwegian. Robert sang the Beach Boy’s “Surfin’ USA.”  His vocal rendition was musically terrible but hilariously funny. In thick accents, the middle-aged Russians next did a spirited duet of the Beatles’ “Back in the USSR.” When they got to the line “those Moscow girls really knock me out,” they rolled their eyes in mock passion. I split a gut laughing. We ended the evening exploring Yokohama like lifelong friends.

Do you want harmonious diversity? An end to nationalism and racism? Private room karaoke may be the answer.

Struggles with patent litigation

A few years later, I acted as an expert witness for HNC in a patent conflict. One of my jobs was to examine HNC code. The software they used predated today’s wealth of available geographical information.  But the code writers cleverly worked around this limitation.

Did you know that you can subtract the zip code of one town from that of another to get a rough estimate of the east-to-west separation of the locations? One of New York City’s zip codes is 10001; Chicago’s is 60007; LA’s is 90001. The further west you go in the continental United States, the higher the zip code number. Before more powerful computers containing more information were available, this was one of the metrics used by legacy HNC software. Credit card transactions occurring close together in time but separated in distance raised one of the many red flags monitored by the HNC credit fraud code.

Postscript

Robert Hecht-Nielsen led a full life marked by remarkable achievements, including founding the world’s first billion-dollar company built on AI. But there is more to tell.

Robert and I were colleagues, but not what I would call close friends. Nevertheless, I received an unexpected phone call from Robert a few years after the Fair Isaac deal. He spoke like we were old friends. Robert had had a near-death experience that transformed him spiritually.

I never shy away from confessing my faith professionally when appropriate, and Robert knew I was a Christian. That’s why he called me. He wanted to share the joy of his spiritual transformation. Robert, now rich and free, was flying his own plane around the world, living a full life in the joy and peace he had found. He passed away in 2019 and would be the first to say that he couldn’t take his millions with him.

Nor, because of his transformative experience, did he care.


The Man Behind the First Billion-Dollar AI Business