Artificial intelligence became a mainstream investing topic after ChatGPT launched in November 2022, but Wall Street had been using computer-driven investment models long before generative AI entered the picture. Today, investors can buy ETFs that use artificial intelligence or machine learning as part of the stock-selection process, putting a version of that technology within reach of ordinary brokerage accounts.
The appeal is obvious. AI can sift through enormous amounts of financial, market, news, and sentiment data and identify patterns that would be difficult for an individual investor to process. But there is a much more important question for anyone putting retirement savings on the line: Does an AI-driven strategy actually produce better returns after fees?
AI Stock Picking Was Here Long Before ChatGPT
The Amplify AI Powered Equity ETF (NYSEARCA) launched in October 2017, nearly five years before ChatGPT arrived. AIEQ uses an index whose stock-selection model runs on IBM Watson and analyzes financial statements, news, economic information, market data, and other inputs. The fund currently carries a 0.75% annual expense ratio. Qraft Technologies followed with the QRAFT AI-Enhanced U.S. Large Cap Momentum ETF (NYSEARCA) in May 2019. AMOM also charges 0.75% and uses AI in an actively managed momentum strategy.
That history matters because AI investing has had enough time to produce more than a few eye-catching months of results. It also shows why investors should be careful with the label. These funds do not all use AI the same way, and the technology does not eliminate traditional investing risks. The models still depend on data, assumptions, portfolio construction, and market conditions. Fund disclosures specifically warn that model or data errors can hurt performance.
The Track Record Is Mixed, Not Magical
AIEQ provides one of the longest clean tests. As of July 31, 2026, Amplify reported a 10.12% year-to-date NAV return, a 15.79% one-year return, a 14.46% annualized three-year return, and a 4.55% annualized five-year return. The S&P 500 returned 10.14% year to date, 19.56% over one year, 19.32% annualized over three years, and 12.86% annualized over five years over the same period. In other words, AIEQ was essentially even with the index in 2026 through July, but it trailed substantially over the longer standardized periods.
That does not mean every AI strategy has lagged. WisdomTree’s U.S. AI Enhanced Value Fund (NYSEARCA) returned 16.06% year to date through July 31, ahead of the S&P 500’s 10.14%, although its 19.35% one-year return was slightly below the index’s 19.56%. There is an important catch when examining AIVL’s longer history: the fund existed before its current AI strategy, which took effect in January 2022. Its 2006 inception date therefore should not be treated as a 20-year AI investing record.
Not Every “AI ETF” Is Actually Picking Stocks With AI
Investors searching for AI ETFs will find funds that own companies involved in artificial intelligence as well as funds that actually use AI in portfolio selection. Those are two very different bets. An ETF that owns semiconductor, software, cloud, or robotics companies gives investors exposure to the AI business. An AI-managed or AI-assisted ETF instead uses a model to help decide which securities belong in the portfolio.
Even within that smaller group, the strategies and costs vary considerably. AIEQ and AMOM currently charge 0.75%. WisdomTree’s AIVL charges 0.38%. FINQ’s new AIUP fund charges 0.70%, while AINT charges 1.25% and uses a dollar-neutral U.S. large-cap strategy that combines long and short positions. Both FINQ funds launched in February 2026, so they do not yet have a meaningful long-term public track record. By comparison, the iShares Core S&P 500 ETF (NYSEARCA) charges 0.03%.
For retirees and long-term investors, that fee gap deserves attention. An active strategy has to overcome its higher costs before its stock-selection advantage reaches the investor. A higher expense ratio does not automatically make a fund unattractive, but investors should demand evidence that the additional cost is buying something worthwhile rather than assuming sophisticated technology will produce superior returns.
What Long-Term Investors Should Take From the Results
AI may become an increasingly useful investing tool, but the public ETF record does not show that giving a machine a larger role in stock selection reliably beats a low-cost broad-market index. Some AI strategies have outperformed during particular periods, while others have lagged. That is not especially different from the broader world of active management, where short stretches of outperformance are much easier to find than a durable advantage across multiple market cycles.
There are practical risks beyond performance, too. QRAFT’s AI-Enhanced U.S. Large Cap ETF (QRFT), which launched alongside AMOM in 2019, was approved for liquidation in July 2026. The SEC filing did not attribute the closure to poor performance or insufficient assets, so investors should not assume a cause. Still, it is a reminder that niche ETFs can disappear, potentially forcing investors to reinvest the proceeds and creating tax consequences in taxable accounts.
For someone building or protecting retirement savings, AI may be most useful as another research tool rather than a reason to abandon basic investing discipline. Before buying an AI-driven ETF, look at the strategy, expenses, liquidity, portfolio concentration, model risks, and performance across several years. The technology may be new. The standard investors should demand from it is not.