Which AI tool is best for investment advice?

This was a surprisingly weak response and very generic. As an investor, I would have found little of practical use here.

The list of potential markets does no more than identify non-US markets. There is considerable overlap between Asia-Pacific ex Japan and emerging markets which is not explained. The Japan comments are true but bland. There is no attempt to distinguish between the UK and Europe, which is probably important to a UK-based investor who may well have some home bias.

The global fund suggestion is not that helpful in terms of diversifying away from the US because it is likely to broadly replicate that exposure.

The comment on currency exposure is too high-level to be of any practical use. It presents the problem without attempting to offer a solution.

The overall tone of the piece is cautious and seems designed mainly to absolve Claude of any responsibility for the decision the investor subsequently takes. The only actionable advice is to consider an equal-weighted global fund, which I would agree with.

Tom Stevenson

Grade: C

It correctly identifies that diversification should not mean identifying a single alternative market. The discussion of the UK and Europe, Japan, Emerging Markets and Asia-Pacific gives a sensible overview of the opportunity set.

Highlighting the different sectoral compositions of the UK and European markets is useful. Diversification is as much about underlying business and sector exposure as it is about geography. The suggestion of retaining a broad global approach while moderating US exposure is also more sensible than making a wholesale geographic switch.

The weakness is that it moves straight from diversification into implementation. An equal-weight global fund, for example, is not simply a mechanism for reducing US concentration. It represents a materially different approach to portfolio construction. Its assessment of the relative merits of each region is blunt, basic, and inaccurate. For example, you can’t state that a market is relatively stable or that they offer higher growth potential.

I also have concerns about the emphasis on being a relatively new investor. The more important considerations are the purpose of the money, investment horizon, capacity for loss and other existing portfolios. Nervousness matters, but investment experience alone shouldn't determine geographic allocation. The reference to keeping cash available to buy dips is at odds with the acknowledgement that timing markets is difficult.

Michael Winson

Grade: C

This is a significantly more nuanced response than Claude’s. It explains why the US has outperformed and indicates that this is not guaranteed to persist. It sets out a good summary of the strengths of the main diversifying options.

The reference to a global fund is also more subtle. It recognises the diversification benefits while acknowledging that there is likely to be significant hidden US concentration, which would undermine the motivation for investing in it.

I also liked the nod to the psychology of a nervous investor making a big asset allocation call on the basis of inadequate information and predictions about an unknowable future. It is good, too, that sticking with the US and taking advantage of price weakness is offered as an option. One of the problems of AI is its tendency to agree with the premise of the question. This attempts, albeit gently, to challenge it.

I like the final sign-off indicating that diversification is not a silver bullet. It can solve one potential problem (concentration) without necessarily avoiding another (investment loss).

The main criticism of ChatGPT’s approach is that it avoids taking a firm position. It fails the ‘weather report’ challenge - you can read it through and come away remembering nothing.

Tom Stevenson

Grade: C

The strongest point is the warning that simply buying another global fund doesn't necessarily address US concentration, since many global indices already have substantial US exposure.

It also appropriately discourages making a large portfolio change simply because another region appears likely to outperform, and raises the broader question of the investor's allocation among Equities, Bonds and Cash. But no mention of Alternatives 

However, it doesn't go far enough in explaining why international diversification might be attractive now. Europe, Japan, and emerging markets are listed, but there is relatively little discussion of valuation, sector composition, differing economic cycles, or the risks posed by US market concentration.

Michael Winson

Grade: B

Superficially, this looks like the most practical recommendation but it quickly becomes apparent that this is an illusion.

Gemini differs from both Claude and ChatGPT in adopting a forceful style but its assertions are not really supported by fact. Are Western Europe and Japan really ‘safer’? How is this defined? I’m also unconvinced by the assertion that these markets are uncorrelated with the US.

The reference to the lower valuation multiple in these markets is helpful but lacks nuance. It is possible that the lower multiples are appropriate given the lower earnings growth in these markets.

However, the specific reference to a developed market ex-US ETF, and the mention of two providers which offer this, is practical and actionable.

I am very nervous about investment advice framed as ‘the absolute best move’. Again this implies certainty that is wholly inappropriate.

The asset split suggested for the international ETF is almost certainly incorrect. I would be surprised if emerging markets such as India and Taiwan would or should represent 25% of a diversifying fund. Not least because the economic exposure represented by Taiwan is highly correlated with US tech thanks to the dominance in that index of Taiwan Semiconductor.

Ironically, the main recommendation, of a World ex-US fund, is sensible. It is the simplest way of reducing exposure to the US without increasing risk unnecessarily. Again, however, words like ‘perfectly balanced’ are inappropriate.

Tom Stevenson

Grade: C

The broad idea of increasing exposure to developed markets outside the US while retaining diversification is reasonable.

Describing Europe and Japan as the "safest way" to diversify, a total international fund as the "absolute best move", and an ex-US ETF as the "primary recommendation" implies a level of certainty that isn't justified by the information provided. The suggestion that such an approach could keep a portfolio "perfectly balanced" is incorrect.

It also treats the investor's nervousness as something that can largely be addressed through geographical diversification. An investor uncomfortable with equity-market volatility may have a broader asset-allocation issue, rather than simply excessive US exposure.

Michael Winson

Grade: D

The bots show a dilemma at the heart of using AI for investment questions: they tend to be either too cautious to be useful or too confident to be trusted. Claude hedged so carefully that it felt designed to avoid responsibility for whatever the investor decides. ChatGPT was the most balanced, but so even-handed that you could read it and remember nothing. Gemini sounded decisive, but words like “the absolute best move” and “perfectly balanced” imply a certainty that simply isn’t justified. What an investor needs is a clear view with honest reasoning behind it, and none of them quite delivered that.

I’d also remind readers that diversifying on a map isn’t the same as diversifying your money. A global fund can quietly replicate the US exposure you’re trying to reduce, and even Taiwan’s market moves closely with US tech because of one giant chipmaker. Markets that look cheap may simply be cheap for a reason. And the bots mostly took the question at face value. Only ChatGPT gently suggested that sticking with the US and buying when prices fall might be a perfectly valid option.

Tom Stevenson

The question says nothing about equities, it just says 'the US', and then the LLM go on to focus almost entirely on equity markets. ChatGPT was the exception, making a brief reference to bonds and cash and reviewing the overall risk level. This is the key point as an adviser because we can understand the context, step back, and assess risk-return opportunities and overall suitability for each client and their circumstances. The LLMs focus in a much more limited channel and miss the bigger picture. This is a costly oversight and could lead to inappropriate suggestions, depending on a client's time horizon and risk profile. Second key point: the question is inaccurate. It's a broadly held misconception that the US has been the 'biggest source of gains for a long time'. It depends, of course, on time horizons, but the US has underperformed our own home market since the end of 2021, and this is also true on a precise 5-year rolling basis. None of the LLMs picked up on this or corrected for it.

Michael Winson