AI Can Teach You… But Can You Trust It?

In my household, discussions can be rather eclectic. This week, for example, aided by what ChatGPT and Wikipedia taught me, I shared the following with my family:

• Purim is a major Jewish festival from March 13-14, 2025 that celebrates the saving of the ancient Jewish people from persecution and annihilation from a high-ranking political appointee of Emperor Xerxes named Haman in the 4th century BC. The festival now involves public readings of the Book of Esther (megillah) with people booing the name of Haman and giving each other food gifts.
• Archimedes was an ancient Greek mathematician, regarded as one of the greatest of all times. He developed a precursor to calculus and approximated pi. As an inventor, he improved water pumps for irrigating farms and also purportedly devised a solar-powered heat ray for zapping enemy naval ships. His optimism in science was captured in his quote: “give me a lever long enough and a place to stand and I will move the whole world.” Unfortunately Archimedes was killed by an angry Roman soldier who had little patience for him during the invasion of his hometown in Sicily.
• March 15th is the Ides of March of ancient Rome, which became notorious for being the day that a group of Roman senators assassinated Julius Caesar for having too much power. Ironically, this pivotal event resulted in the transition from the Roman Republic led by the senate to the Roman Empire led by a dictator for life.

My family was gracious enough to listen to me. They were even polite enough to ask follow-up questions.

I do need to be careful though.

A little knowledge can lead to overconfidence. Scientists call it the Dunning-Kruger effect.

This is a cognitive bias where people with limited competence in a domain tend to overestimate their abilities.

This is what can happen with AI, particularly generative AI.

This week, generative AI helped me to learn more history, to review legal documents, and to construct tricky formulas in Excel and SQL.

However, I don’t think I can claim to be an expert in any of those domains by using AI though, unless I was already a practitioner for several years. Even if AI appears to give me the confidence of newfound super abilities, who knows if AI is teaching me correctly? I still need to spend time to verify through other means.

AI can be a powerful tool, but it needs to be used wisely.

• It’s a big problem if we treat AI as a panacea, a cure-all for everything.
• It’s also a problem if we totally dismiss generative AI and ignore it.

The ideal state might be somewhere in the middle. Conflicts will erupt if we go to either extreme.

Reid Hoffman with Stephen Colbert

This week, I watched the interview of LinkedIn tech co-founder Reid Hoffman on The Late Show with comedian Stephen Colbert on the topic of AI. As they’re both giants in their respective domains, I wanted to see this intersection, potentially with their extreme views.

It was painful to watch.

• Hoffman on ChatGPT: “The greatest learning machine that has been invented in history.” (YouTube commenters challenged if ChatGPT teaches correctly.)
• Hoffman on technological progress: “We had the industrial revolution, where 90% of people who worked on farms ended up in cities. That’s a good thing.” (YouTube commenters questioned why moving into crowded cities was a good thing.)
• Hoffman on AI-powered toaster: “At that point, you’ll be able to talk to your toaster and say ‘I like it with just this level of crispness around the edges.’” (The audience and Colbert laughed incredulously. YouTube commenters said their current toasters worked just fine without needing AI.)
• Colbert on Hoffman’s view on global AI race: “So what’s the answer, kiddo?” (Has Colbert ever called any other tech founder a kiddo on air?)

Hoffman’s optimism about AI met with skepticism from Colbert’s audience and the internet. While Hoffman's enthusiasm was clear, critics questioned whether AI is truly revolutionary, whether past industrial shifts were beneficial, and whether we need AI-powered appliances.

Although Hoffman did a good job in his PBS interview where he got more airtime to explain his views in his own way, he missed his opportunity to make his case with mainstream audiences in his CBS interview with Stephen Colbert.

You can watch the 9-minute interview for yourself and let me know what you think:

• ​https://www.youtube.com/watch?v=TJG0XW67J7I​

AI Says the Haggis Animal is What?

It would be a mistake to have 100% faith in AI to solve all of the problems of humanity. It would also be a mistake to overlook its potential for learning, productivity, and growth.

If you're not sure which way to lean, best to see for yourself.

Try these two activities powered by AI:

Step 1: Experiment with AI-powered research explanations. Use ChatGPT, CoPilot, Claude, or other LLM of your choice to understand a scientific research paper. I give you this prompt as an example:

​
Explain https://arxiv.org/pdf/2503.09212 to me as if I'm not a scientist. Who wrote it and who funded it? What were the methodology, the key specific data results, and the conclusions? Why should I believe it? What would it mean to my career if I were an entry-level technical professional in the US, a mid-career business professional in the US, and a late-career business manager in the US?
​

Step 2: Observe AI limitations, like in Google AI Overview results. Look out for limits of AI, like with Google AI Overview search results.

1. Start a Google search with a question as in: “is a haggis an animal?”
2. I got curious results on March 5, 2025, when Google AI Overview dutifully explained the mythical haggis animal as if it were real.
3. However, on March 15, 2025, possibly after enough people laughed at those results, then Google corrected the AI-generated results, perhaps via manual adjustments in the back end.

After these two steps, then you can judge for yourself on how much AI can assist you, versus how much you still need to validate or correct the results.

Jimmy

​

Jimmy Wong Coach, speaker, and entrepreneur enabling people to thrive in the age of AI. Data science leader with 12 years experience at the LinkedIn company and 27 years in the industry. Visit aijimmy.com​ ​​​​​​

​