By Jack Cumming

How much time have you spent trying to understand what artificial intelligence (AI) is and how it works? I’m guessing the answer is “lots of time.” I’m going to guess further that you think you are wise, but you don’t connect AI with wisdom. In fact, at first, I thought of AI as no more than a machine that thinks- an old idea- and machines aren’t wise. Then I stumbled across a 14-minute presentation on YouTube that provided more insight than everything else.

Chloe Lubinski

The speaker is Chloe Lubinski, who self-describes her role as leading “research partnerships with the world’s wisdom traditions.” She works for Anthropic. Anthropic’s product is an AI model called “Claude.” The Anthropic group broke away from OpenAI and its “ChatGPT” product to form a separate corporation to allow itself to be more principled and less commercial.

You may remember the confusing weekend when the controversy broke out into public view. It’s not surprising that Chloe Lubinski has a job (1) to explain AI to those who struggle to understand it, and (2) to connect AI with wisdom and human understanding. That mirrors the qualms that took Anthropic’s founders away from OpenAI. It mirrors, too, the complementary talents of the brother-sister team who lead Anthropic.

The PowerPoint Distraction

There’s another, less world-changing, bit of wisdom embedded in the 14-minute video. Ms. Lubinski evidently used PowerPoints in her talk. Doesn’t everyone today? As you concentrate intensely in order to grasp everything that she is sharing, think of how distracting it would be if there were slides to be read and absorbed while you tried to focus on her message.

There is a place for PowerPoints. They’re good with pictures and graphs. They’re terrible with words, which is what they usually share. The rationale is that there are people who learn from listening, others who learn from reading, and those who want a quick list-style summary. If you’re a thought leader, you probably concentrate on the speaker and not the slides.

Chloe Lubinski should give herself more credit. She doesn’t need the slides. Her message is more powerful without them. I would love to have heard the give-and-take with the audience. That wasn’t included, but just the 14 minutes we have will help you to better understand AI and how to make sure that any deployment of AI is constructive within your organization and life.

Tech Crazes, like AI, Are Not New

Listening to Ms. Lubinski took me back to the late 1960s and my personal experience then as a young actuary who loved progress. My task was to develop a set of rates for disability income insurance, and I wrote a computer program to replace the calculations clerks had been doing up to that point. We had a brand-new capability, Call360, offered by IBM, that allowed easy access to its computers for analytical work.

My immediate boss was enchanted. He had me run scenario after scenario, each with different assumptions, outcomes, and likelihood. That would have been impossible previously, when two clerks, a doer and a checker, would have taken a week to complete a single scenario that the computer could complete in ten minutes. The final stage was to get clearance from the Chief Actuary in his imperial office high up on the 37th floor of the building.

Visit to the C-Suite

In the 1960s, this was the culminating step. My boss and I went to the meeting with trepidation and machine printouts. We were met by the courtiers, mostly professional women executive assistants, who attended to the needs of high executives in that majestic era of long ago. We were seated in a conference room adjacent to the chief’s quarters.

Finally, the imperial presence, himself, swept into the room, brought the meeting to order, and then turned to my boss to start his presentation. “Mr. [Chief],” my boss began, “these results have been prepared by computer,” and he swung a printout in front of the chief. That was enough to sway the Chief Actuary, who may have been reluctant to show his ignorance of how it was possible for a computer to do anything other than mass recordkeeping.

Parallels Today

My sense is that that is where many executives may be today when it comes to evaluating AI claims. Every IT vendor seems to open each sales presentation by claiming how adept they are at embedding AI into their offerings. Some may be meaningful, like Epic’s use of AI to enable physicians to maintain eye contact with their patients, but other claims are little more than irrelevant nonsense. It takes a wise executive to know the difference.

If you can spare 14 minutes, use them to watch Chloe Lubinski. It will redirect your thinking about AI in a constructive direction, just as Ed Berkeley’s 1950 book, “Giant Brains; Or, Machines That Think,” did in an earlier era. I don’t think our Chief Actuary had read it, even though Berkeley was also an actuary and a friend of the chief.

Afterthought

We hear of three main models for everyday AI usage by ordinary people: “ChatGPT”, “Claude”, and “Grok.”

  • So far, OpenAI’s ChatGPT has a reputation for commercialization. Its distinction is that it was, first and foremost, controversial for having opened AI to the public on November 30, 2022. It’s led by Sam Altman.
  • Anthropic’s “Claude” has a reputation as a conscience-motivated breakaway from OpenAI. The company’s signature is safety. It’s led by a brother and sister, with the brother being highly technical and the sister leading the human side of the organization.
  • “Grok” is the brainchild of Elon Musk. He, too, broke with OpenAI in reaction to its commercialization. As with other Elon Musk enterprises, Grok is known for its engineering prowess.

Of course, there are others – DeepSeek is very popular in China, and Google has Gemini embedded in its search – but depending on how you view your own personality and character, you may be drawn to one model or another. Amazon is struggling to catch up with “Alexa +,” and Microsoft depends on OpenAI. This article focuses primarily on Claude. Click here for a longer Bloomberg Originals interview exploring Anthropic’s approach and its leaders.

Each model is like an individual. Each has its own character. Each has its own abilities. Just as people reflect their parentage and upbringing, with some people having more integrity and character, so each model reflects its parentage and training.

People inherit genes from their parents. AI models are stamped by the people who initiate them.

People reflect values learned as children, from their upbringing. AI models reflect the values they are trained on.

Thus, AI isn’t one thing. It’s as varied as human nature is. If it’s trained to be mean, it’s likely to seem mean-spirited. If it’s trained to be exemplary, then it’s likely to show integrity.

For CEOs grappling with AI, evaluating it can be like finding human talent. The quest for talent is the most important job of a CEO. Choosing an AI approach is also a talent search.