- Sign up for a free subscription at larrykotlikoff.substack.com. *I’ll immediately make it a free lifetime subscription.
David Blanchette is Prudential’s Head of Retirement Research. He’s a widely respected retirement-planning guru (and a past guest on Economics Matters). In 2014, David coined the term *Retirement Smile *in an article published in the Journal of Financial Planning . Retirement Smile references arranging your retirement spending to spend more during your Go-Go years, spend less during your Go-Slow years, and spend more during your No-Go years.
Go-Go years refer to your early retirement when you want to travel abroad, eat in fancy restaurants, hire a private trainer, party in Vegas, ski Aspen, golf at the best courses, shop till you drop, etc.
Slow-Go years are mid retirement — when you’re sick and tired of getting sick and tired on cruise ships, can’t stomach the mounds of salt they call food at high-priced restaurants, and are routinely hitting the ball into the rough.
No go years are no fun years. It’s when you can’t make it up the stairs, let alone fly to Bali, your best friend is your GP, and your out-of-pocket medical bills are giving your ulcer an ulcer.
Not every retiree wants to build a spending smile. But those who do need help to ensure they don’t end up with a frown. Take single, childless, 60-year-old Carol, who lives in DC. Carol loves her $2K a month rental apartment, which she plans to occupy for the rest of her days. She expects her rent to stay even with inflation.
Carol retired last year after a great career. Her $40K salary at age 20 increased annually at 3 percent. Unfortunately, her company didn’t have a retirement plan. But Carol socked away $1 million in regular assets. She intends to take Social Security at 70 to maximize her monthly benefit.
Carol read the Bankrate article that named MaxiFi Planner “Best Financial Planning Software of 2025.” She ran it leaving her maximum age-of-life setting at its default — age 100. Carol sees inflation creeping up. She set the program’s inflation rate at 3 percent and her nominal return at 5 percent. This translates into a 2 percent real return, which, given prevailing rates, Carol can achieve by buying a TIPS ladder that funds MaxiFi’s calculated annual asset withdrawals. (This recent podcast with Kevin Esler shows you how to use Kevin’s fab and free software tool — tipsladder.com — to create your ladder based on MaxiFi’s specs.)
My buddy, Rob Berger, the highly influential financial-planning-software influencer, has repeatedly suggested that MaxiFi requires households to have a perfectly smooth living standard. This has never been the case.
No worries, Rob. Please look at the Living Standard Index settings at the end of his blog that Carol specified to make her smile. Carol accessed this index by first clicking Settings and Assumptions in MaxiFi’s input menu and then clicking on the tab called Living Standard. The index’s current-year value is always fixed at 100. Carol specified a 1.8 percent annual decline in the index through age 80. Then she grew the index from its age-80 value of 70 by 1.8 percent annually. This produced an index value of 100 at age 100.
Carol ran MaxiFi with the above-specified inputs and, voila, out came her discretionary-spending retirement smile shown in green bars in the top chart. Carol’s 2026 discretionary spending is $53,923, declining to $37,746 at age 80 and growing to $53,923 at age 100. (All figures are in 2026 dollars.)
Gemini, to which I fed all the data and explained the definition of discretionary spending, calculates Carol’s 2026 discretionary spending at $47,150 at age 60, $32,788 at age 80, and $46,845 at age 100. Thus, Gemini, into which Google has poured hundreds of billions of dollars, generated a smile that is 13 percent too low in each of Carol’s potential remaining 40 years! To generate Gemini’s result in MaxiFi, Carol would need to enter not her $1 million in regular assets, but $818,100. In effect, Gemini is depriving Carol of the use of 18.2 percent of her assets!
What is Gemini getting wrong? That seems like a good question. But it’s the wrong question. It suggests that tweaks to AI will magically permit AI to use the wrong algorithms to correctly solve a highly complex, non-linear mathematical problem — a problem that can only be solved with the right algorithms applied in the right order. By analogy, the question amounts to asking how to tweak trigonometry to solve a problem whose solution requires calculus.
Nowhere in Gemini’s answer does it say, “Here’s my best guess.” Instead, when I asked Gemini how sure it was of its answer, it replied, partly in bold, “I'm very confident in the mathematical logic of the model.”
Let me say, I’m very confident from this as well as this, this, this, this, and this Substack blog that AI doesn’t have a clue when it comes to providing accurate financial advice. Who knows if it ever will. But, then again, why should Carol care? She can license MaxiFi and get the correct answer in three seconds — an answer whose validity she can immediately verify with her eyeballs. In particular, she can view MaxiFi’s lifetime budget report and see that her lifetime resources exactly equal her lifetime discretionary spending plus her fixed spending on housing, taxes, and Medicare IRMAA premiums.
I’m harping on the use of AI for a reason. Many of you, now 137,000-strong subscribers, may be asking AI for financial advice. Don’t! It pretends to make proper mathematical calculations when it can’t. What it’s doing is predicting which algorithms to apply and in what order. But it’s drawing on known algorithms when the right algorithms aren’t in the public domain. As for just predicting the answer, i.e., not trying to calculate it, no one in any of the AI companies has asked me to train their LLMs on MaxiFi’s results. Such training would be straightforward based on billions of cases we could easily construct by perturbating observations in the Federal Reserve’s Survey of Consumer Finances - observations we could then run through MaxiFi.
Let’s modify Carol’s situation slightly. Suppose she has $1 million in assets, but half are in regular assets and half are in an IRA. Further assume that Carol has decided to start her IRA withdrawals at age 75 when her RMDs (Required Minimum Distributions) kick in. Here’s what happens to Carol’s smile. Kind of toothy, eh?
What’s going on? I didn’t change Carol’s Standard of Living Index smile values. I simply reduced her regular assets by $500K, added a $500K IRA, and set her age of smooth IRA withdrawals to age 75.
Carol’s decision to delay IRA withdrawals until age 75 makes her cash-flow constrained. MaxiFi accommodates desired living-standard profiles, but not if doing so puts you in debt. (Note to Rob: You can override this no-borrowing default, again, under Settings and Assumptions by clicking Other, setting and specify the interest rate at which you can and wish to borrow.)
As you can see, Carol’s living standard falls through age 70, jumps a bit at age 70 when she starts Social Security, falls through age 75 and then jumps up at 75 when her IRA withdrawals kick in. After that, MaxiFi complies with her desired smile by reducing spending through age 80 and increasing it thereafter. Thanks to her cash flow problems, Carol’s age-discretionary spending profile comprises two half smiles followed by an asymmetric full smile.
Can Carol regain her full smile without seeing the dentist? Yes, she can simply tell MaxiFi to start her IRA withdrawals immediately rather than at age 75 (Rob, click *Settings and Assumptions *and then Retirement Accounts. Here’s Carol’s new smile.
Surprisingly, starting her IRA withdrawals early doesn’t cost Carol anything. On the contrary, Carol’s lifetime discretionary spending rises — from $1,224,555 to $1,281,220! Yes, waiting to start withdrawals until 75 lets Carol’s IRA accumulate asset income tax free for longer. But Carol’s post age-75 federal, DC, and IRMAA taxes are now considerably higher than if she starts her IRA withdrawals at 60. Intuitively, bunching her withdrawals over 25 years starting at age 75 — and, thereby, landing in higher federal, DC, and IRMAA tax brackets, is worse than spreading her tax liabilities over the next 40 years. All told, in withdrawing from her IRA early, Carol lowers her lifetime taxes by $56,665!
I entered all of Carol’s inputs that produced the failed smile chart. In particular, I told ChatGPT that Carol has $500K in regular assets and $500K in an IRA, that she would start her IRA withdrawals at 75, that she couldn’t borrow, and that she wanted to achieve the above-described discretionary spending smile.
ChatGP said Carol should spend $68,000 this year and $43,000 at age 75. The correct answers are $$34,070 this year and $49,549 at age 75. I asked ChatGPT its confidence in its answers. It responded, “I’m highly confident numerically.” Nice to know.
I could check the other AIs. But there are dozens of them and, from past experience, their answers will be equally awful and very different one from the other.
First, MaxiFi is the only tool that you can use to make a retirement smile. You can use its Standard of Living Index to do so. Alternatively, you can enter age-specific special expenses. Examples of special expenses would be a cruise every year for the first 10 years of retirement costing $20K per year and a nursing home stay between ages 90 and 100 costing $100K a year. In this case, your fixed spending, which includes these special expenditures, will incorporate these smile features whereas your discretionary spending will remain smooth assuming a) you leave its index values at 100 (Note: This year’s value is, in any case, always fixed at 100.) and b) you aren’t cash-flow constrained.
Second, if you are compelled to ask AI for financial advice and follow that advice, make sure to take a screen shot of what it tells you. Also, document that you did what it said. This way, you’ll be able to join the class action suits that are sure to arise.
Third, the biggest concern with building a retirement smile is running into a cash-flow constraint. But you can alleviate cash flow constraints by withdrawing from retirement accounts early, working longer, downsizing your home, moving to a low-tax state, delaying Roth conversions, allocating fewer regular assets as reserve funds that aren’t to be spent, and, if need be, taking Social Security early. MaxiFi lets you explore all of these options, one by one or in unison.