Michael Heekin first took the Club on a nostalgic trip through the history of the internet featuring names such as America Online, the first internet connection many people had, Alta Vista and Yahoo, early search engines, and Netscape, an early browser.
He discussed how electrical energy demands from Artificial Intelligence (AI), various data centers, electric vehicles, among others will be huge. Previous Club speakers have addressed the energy needs as well.
“AI does not think like the human brain,” Heekin said. “Data is the key. It does not have instincts, emotions or consciousness.”
He said that as far as we know, AI is not capable of innovative thinking. He acknowledged that while AI might replace humans in performing some tasks, it’s power will be helping humans work more efficiently.
“It can take the works of the Beatles, and it can compose a new song and have Paul McCartney sing it,” he said. “It sounds like the Beatles because it’s got the same meter and the same interplay of instruments. But, it’s not capable of, and I’m going to guess never will have original thought. It’s not going to recreate a masterpiece like the complete works of Shakespeare or the City of God by Augustine.”
Heekin, who specializes in the management and development of growth-stage health and technology companies, talked about how we saw internet companies succeed and fail, and through the booms and busts, we’ve seen what they needed to succeed. AI companies will be similar in that there will be successes as well as dashed dreams and lost investment money.
“Access to good data and good data curation is what’s going to be really key,” he said. “Getting good data, curating it, and then stuffing it into the algorithm.”
He said that regulatory good faith is necessary for AI to grow, and he praised President Bill Clinton and U.S. House Speaker Newt Gingrich for their light regulation of the Internet in the 1990s.
“The government needs to not kill this thing in the crib. It has great potential for good, and great, great potential for threats. But that’s the same with a lot of things, wi-fi, life in general. The genie’s out of the bottle. If it’s not AI, it will be something else.”
He also looked at social media and regulation as a model for AI.
“Social media is like giving kids a pack of cigarettes. It’s clear AI is going to need some regulation, but it should be ex-post, not in advance. People in Washington don’t have a monopoly on good sense.”
He said AI will be a commodity product.
“There will be some wrinkles to it, but it will be a commodity. We will have access to AI, but most AI companies will not own the AI algorithms. Thay may own developments of it. They may have prompt engineers that can tell the AI algorithm what to do, but they will not own the algorithm lock stock and barrel.”
The public domain is very valuable to AI’s ability to learn, but so is input from users.
“There’s a dirty little secret to ChatGPT,” he shared. You ask it a question, and it’s learning at your expense. It’s learning from what you want to know. AI uses all the public domain data it can, but it also acquires data from data moats that are filled with consumer behavior data just like what Publix uses.”
As for valuation of AI companies, Heekin again pointed to the early Internet. When companies presented their ideas on PowerPoint slides, Wall Street gave them valuation. When the companies started making money, the valuation could change.
“When they started getting revenue,” he said, “Wall Street started valuing companies at twelve times revenue. If you’re making $1,000, $12,000 is what your company is worth even if Wall Street said it was worth $12-million last month.”
He predicted the same dynamic will work with AI companies. It is incumbent on AI companies, and investors, he said, to recognize that and be smart about their deal making activities.
(You can also view the entire Club meeting on YouTube.)