Why You Shouldn't Take AI's Financial Advice at Face Value

When you talk to artificial intelligence you get the impression that it’s thinking. That its weighing the pros and the cons, and really trying to come up with the best possible answer for you, the way a good friend or relative would. And because of how well it answers difficult fact based questions (e.g. how does a nuclear reactor work), it’s easy to think the answer it gives for your hypothetical and multi-faceted problem must be the right one. But AI isn’t actually thinking, and, even though it has access to all the information in the world, it doesn’t mean it’s capable of generating solutions to what if scenarios.

I mean it is, but explaining how a nuclear reactor works, summarizing the 5 best exercises for achilles tendon rehab, or writing a business plan, is not the same as solving a real life problem that requires weighing multiple possibilities. Everyone has a really smart friend that they would never ask for practical advice from. Because they, like AI, are more robot than human.

I recently decided to roll over my 401k into an IRA after reading an article by the Financial Samurai outlining all the pros and cons of making the switch. I posed the question, ‘should I roll over my 401k,’ to Claude to double check it was the right move and see if it could pull anything else I might have missed. Claude confirmed it was the right choice (pretty straightforward), and re-iterated a lot of what was in the article (presumably it even pulled from it). Which felt reassuring.

I than asked Claude to analyze my current 401k allocations, and make recommendations for where to invest my money in the IRA to most closely resemble the allocations of my 401k. And it did. Claude was able to provide a great analysis of the funds my money was in, the cost of those funds, ETFs to put my money in under the IRA, and estimated cost savings over a 20 year horizon (the time in which I would be legally eligible to access my IRA without penalty) by making the switch. Which again was helpful in confirming it was the right decision.

But things got a little hairy, to me at least, when I asked for advise on timing of the purchases of the ETFs. The money would arrive in my IRA in cash, and would sit there until I bought stocks and put the money back to work. I asked ‘the market is at all time highs, should I wait and spread out my purchases, or go all in with a one time lump sum?’ The response was measured and based on historical findings, and summarized below.

“I'd lean toward lump sum investing immediately upon rollover, or at most a 3-month Dollar Cost Averaging (DCA) if the psychological comfort of averaging in is worth it to you. The cost of waiting — in both missed growth and mental overhead of trying to time the market — almost certainly outweighs the benefit over a 20 year period. The worst investing decision at your age isn't buying at a high. It's sitting in cash waiting for a perfect entry point that may never come.”

Which is good and sound advise for what I would call ‘normal’ times, but I don’t think that’s what we’re in right now. We’re in the middle of a war that has totally disrupted global energy supplies, and dramatically increased the cost of oil, which will dramatically increase the cost of everything that relies on oil to be produced. We’re just a couple of months away from perhaps the most important mid-term elections in history. We’re at odds with all of our allies, and threatening to invade other countries around the world. And, somehow, markets are still at all time highs. So, I posed that to Claude, and got a revised response.

“Revised recommendation: DCA over 3-4 months rather than lump sum. Not because you can predict the market, but because the specific risk profile here — unresolved geopolitical conflict with identifiable downside catalysts — is meaningfully different from a normal "markets are high" situation. Spreading across 3-4 monthly investments gives you some exposure immediately while managing the risk of a ceasefire collapse. That said — do not wait indefinitely. The worst outcome is sitting in cash for 12+ months while the market continues climbing on AI momentum despite the war.”

While not dramatically different, and the advise to not sit on the cash forever remained unchanged, it became obvious to me that in the first recommendation the war and other important factors were not actually considered. The recommendation was based on historical findings, which may prove correct, and perhaps I was looking for a specific answer that I didn’t get, but it was eye opening to me. Because when I talk to AI I assume it is taking every single ounce of information, new and old, relevant and indifferent, into account when it spits out an answer. I assume it’s thinking of all the possibilities that I am. But that’s not always the case, and I think that’s important to know..

I realized that as powerful as AI is. As quick as AI is. As helpful as AI is. We need to still understand that AI is not capable of weighing every possible scenario. That, at least in their current iterations, AI is strongest when it comes to clearly defined fact based missions, and that we need to push back and make sure the answers were getting have taken everything we know to be important into account.

But who knows, 12 months from now I might be sitting typing a new post. “Why I should have listened to Claude the first time. How I lost out on 12 months of a bull market.”