We are approaching the time of year – late autumn – when the Atlantic salmon swim to the gravel beds at the top of UK rivers to spawn. Successful fly fishing is basically a matter of capturing their attention on the way. 

You cast a shiny, sparkling fly and wait patiently, hoping the salmonoids will rush excitedly towards it. Curiously, they are no longer even seeking a meal at this point in their lives – they stop feeding when they enter freshwater. Nevertheless, some cannot help but chase the fly… and get hooked. Only those that stick to their mission reach the breeding grounds.

The parallel might sound incongruous, but we believe a similar story often plays out in the investment world. Confronted by something unusually exciting and attractive, many investors just cannot help but take the bait.

We have witnessed this phenomenon recently. Artificial intelligence has emerged as one of the most popular investment themes of recent years.

There is much speculation that AI will prove to be a source of enormous disruption and opportunity, particularly over the longer term. Many commentators believe it will transform working practices, drive productivity and underpin a growing array of innovations and applications in almost every sphere imaginable.

But we think it is by no means a given that the businesses seen to be at the heart of this revolution today will be the major players several years from now. 

Bubbling under

The dot-com bubble was a lesson in why investors should not get too carried away by a theme. Most of the ‘darling’ internet companies at the beginning of the internet revolution failed to become the long-term winners that shareholders hoped for.

We know that the hypothesis which fuelled the boom – essentially, that the future would lie in ‘clicks, not bricks’ – was sound, yet many of the businesses involved were anything but.

The cast was comprised of a remarkable mix of pioneers, makeweights, bandwagoners and charlatans. The subsequent casualty list was lengthy, with even a number of erstwhile trailblazers adding to the tally.

Take AltaVista. The company was responsible for one of the earliest successful search engines, but it soon lost ground to its rivals – Google being foremost among them – and was sold to Yahoo in 2003. What little remained of its service was quietly shut down in 2013.

Other high-profile failures included eToys.com and Terra Networks. The former racked up debts of almost $250m (£198m), while the latter’s share price careened from $11 to $158 to $3 within a few tumultuous weeks.

Between January 1995 and March 2000, when the bubble was inflating, the Nasdaq went up by 582%. Between March 2000 and October 2002, when the bubble finally popped, the index plummeted by 75%, wiping out most of the gains accumulated during the preceding period.

There may be a catch…

We believe there is a big difference between most of those tech-bubble stocks and mature proven businesses like Microsoft, Apple and Nvidia, which have enjoyed big share price hikes over the past 18 months. But it would still be foolish to ignore the warnings of history that all empires fall eventually.

Alphabet has invested more than $12bn (£9.5bn) in AI, but how soon, if ever, will this translate into strong returns for investors? And will other big investors be able to monetise the benefits that AI can bring to their offerings?

Enthusiasm around AI has arguably fuelled much of the recent rise in the Magnificent Seven stocks, but looking further afield, we believe there are many smaller companies with the potential to deliver more enduring returns from AI.

One example is VAT Group, a leading maker of the vacuum valves vital to the manufacture of semiconductors. It enjoys a market share of about 75%, and we believe it is unlikely to surrender its dominance to inferior producers. Relx (REL), which uses AI and its proprietary databanks to support multiple professions, is another company with a genuine edge, in our opinion.

We believe discernment and prudence tend to be useful attributes in the face of anything resembling an investment craze. Even if it somehow turns out as billed, many of the constituents of ‘the next big thing’ are unlikely to succeed over an extended timeframe.   

It is also worth remembering, of course, that investments of equal or even superior potential can still be found in other arenas. Recent turmoil in the realm of technology has again underlined the wisdom of constructing diversified, balanced portfolios.

For many investors, AI is undoubtedly enthralling and eye-catching. Investors – like those hapless, bedazzled salmon – could be forgiven for having a nibble at something so tempting. But falling for its charms hook, line and sinker may be a bite too far.

Louis Florentin-Lee is a portfolio manager for the Mid Wynd International (MWY ) investment trust.

Any opinions expressed by Citywire, its staff or columnists do not constitute a personal recommendation to you to buy, sell, underwrite or subscribe for any particular investment and should not be relied upon when making (or refraining from making) any investment decisions. In particular, the information and opinions provided by Citywire do not take into account people’s personal circumstances.