Is “Jagged Intelligence” The Only Thing That Is Going to Save the Human Race From AI?

As a retired lawyer, and a judge for twenty years on the bench, I keep hearing that we won’t have any need for lawyers anymore in the future because AI will be able to write legal documents and AI can help someone act on their own in the courthouse.

Having been a family court judge when 80% of the cases in front of me were self- represented, I can tell you it was a nightmare for the litigants representing themselves and a nightmare for me as the judge. And that was before AI.

All the nuances and interpretations of the law and rules and regulations were beyond someone who had not gone to law school and practiced law. They could recite what they read and heard, but they couldn’t apply the law to their real circumstances and analyze and expound on what was applicable to their situation alone.

The Florida Family Law Bar and bench tried to set up systems to assist those representing themselves, but filling in the blanks on a form is different than being able to apply what they wrote and then argue before a court as to why you’re entitled to relief or why the other person is not entitled to relief.

I recently learned that there is a new administrative order in my circuit which requires both lawyers and self-represented parties to acknowledge and reveal that they have used AI as assistance in research and their pleadings. I heard that AI has what is called in a “hallucination problem,” where it wants to please you so it makes things up to make you happy, and then in the court system AI has been making up cases and results that are good for your particular matter before the court that are not real. Sounds crazy, but it can cause havoc in the court system as the judge relies on the cases cited for its principles to follow in making rulings.

So, when I finally read an article that made sense to me, “Jagged Intelligence Can Reframe the Debate over AI Smarts,“ New York Times, April 17, 2026, headlining the front page of the Business Section, by Cade Metz, I was ecstatic someone was finally addressing my skepticism.

I found this article very comforting. Especially the paragraph that says “[t]he Internet holds only a fraction of human knowledge. It records what people do in the digital world, but contains comparatively little data about what happens in the physical world. That means these AI systems can write emails, answer questions, riff on almost any topic and generate computer code. But because AI systems reproduce the patterns they find in digital data, they are not good at planning ahead, generating new ideas or tackling tests they have not seen before. AI does not have general intelligence, what it has is a lot of different skills.”  May I repeat, I agree and cannot imagine they can analyze, apply the knowledge and create anew as a lawyer. . . or a judge.

The author, Cade Metz, introduces us to Andrej Karpathy, a founding researcher at OpenAI, a former head of self-driving technology at Tesla, and, on social media, one of the most closely watched commentators on AI. He was the one who coined the term, “jagged intelligence.”

Andrej Karpathy is quoted to have written, in 2024, that with AI “some things work extremely well, (by human standards), while some things fail catastrophically (again by human standards), and it’s not always obvious, which is which,” comparing AI to the human brain where we learn from birth to adulthood and knowledge and problem-solving capabilities improve together.  I had to admit that I stopped reading the rest of the article at this point and found Andrej Karpathy on Instagram and now follow him.  I am beginning to understand AI. I want to follow what I like.

Then I come to the end of this long article and the author’s conclusion that the wild card is that AI is quickly improving, and that many of the weaknesses identified by Andrej Karpathy and others in 2024 and early 2025 are no longer there. Cade Metz, the author, instead is in the camp with Alex Imas, who is quoted extensively in the article as well, who is an economist at the University of Chicago’s Booth School of Business.  Alex Imas is quoting as saying AI has a lot of different skills that are showing incredible improvements. Alex Imas is the one who gets the last word at the end of the article, “the valleys in the technology are closing.”

I’m so glad the author didn’t quote at the end of the article what was in the middle of the article, “[s]o far, outside of the technology industry, there is only anecdotal evidence that AI has become a job killer. But given how quickly the technology is improving, many tech experts argue that whether AI replaces other kinds of white collar workers is not a question of if, but when.” I guess the author just kept the best quote for last, but the concern about AI replacing real humans at real jobs is what lasts.

There are certain times now where I’m glad I’m retired but then I remember that I have six grandchildren and I want there to be real careers still existing for them in the future.

Is the only thing that is going to save their future “jagged intelligence,” the possible limitations that cannot be overcome despite how optimistic the AI researchers are?

I want their futures to be filled with

Joy,

Mema