Artificial Intelligence - the end?

Is Artificial Intelligence really a game changer??

Arnie’s terse request for an ‘Uzi 9mm’(1) is just a tiny part of a long-running fictional case against Artificial Intelligence (AI). Much of this fiction riffs on the idea that an alternative intelligence would soon decide that humans are superfluous or incompatible with happiness, planetary survival, or more besides.

The sooner we have machines making machines, capable of recursive self-improvement, the sooner we will usher in our (mass) redundancy. That post-human future is already in our midst according to some, with the starting gun fired by the stunning leap embodied in the latest batch of generative Artificial Intelligence. This week, we have a look at where this might lead.

Superpower

New technology is at the heart of humankind’s success over the millennia. For the last 200-odd years in particular, the arrival and incorporation of new technology has accelerated dramatically. One famous economic historian argues that there has been as much technological change in the period from 1870 to the present day as there was between 6000 BC and 1870. (2)

That period of often breath-taking technological change has been matched by living standards too – a doubling of life expectancy (Figure 1), the decimation of absolute poverty (Figure 2), and more besides. Technology has gifted us superpowers, even if they quickly appear humdrum once everyone has them.

Nonetheless, measured productivity growth these last few decades has been quiet, even if the pace of technological change has still seemed dramatic. In 2021, the average American spent eight hours a day online, more than double the amount seen in 2011. Part of that change has been an information revolution unlike anything since the explosive arrival of the printing press in Europe. Then, as now, the ensuing jostle of ideas has created societal and political turbulence.

Into this combustible context comes generative Artificial Intelligence. While traditional AI has been helpful in making predictions, Generative AI is about generating content such as text, video, images, or computer code. One provocative excerpt from a competitor’s related analysis suggested the equivalent of 300 million full-time jobs could be exposed to automation. (3)

Why is this different?

The generative AI technologies currently in focus are distinguished by three main characteristics: First, the generalised rather than specialised use cases. Second, their ability to generate novel output (easy to mistake for human) rather than merely describe or interpret existing information. Third, their approachable interfaces that both understand and respond with natural language, images, audio, and video. The first two advances are potentially key to expanding the set of tasks that AI can perform, while the third is already central in this technology’s rapid adoption. (4)

What could go wrong?

No new technology has proved only positive in the past. All have come with difficulties, disruption, and worse. We should expect the same of Generative AI and its more generally aware potential progeny. (5) The printing press mentioned above can be linked directly to trends in soaring European literacy and ultimately the scientific revolution, one of the vital building blocks for subsequent European economic take-off in the 17th and 18th centuries.

However, the press also sparked the wars of religion and decades of bloodshed. Without the press, we may never have seen the cancerous spread of some toxic ideas and ideologies. It is obviously hard to imagine the counterfactual world where a particular innovation did not emerge. Nonetheless, most of the time, with a bit of time and perspective, we can discern clear net benefit from the arrival of new technologies.

Unfortunately, that is likely as specific as we can plausibly get. There is a predictable clamour amongst investors and frankly everyone to see where this technology takes us. Will it, as one critic argued, unhelpfully drive the cost of nonsense production to zero? Or could it be the next General-Purpose Technology (GPT) we’ve all been waiting for this last few decades, joining the ranks of electricity, the microchip, the compass and other game changers?

It could of course do both and much more besides. If we go back to any of those technological breakthroughs and take the associated temperature of the commentariat, you see very little prescience. Understandably so. Even gauging the ripples from the most mundane breakthroughs of the past have proved beyond contemporary soothsayers.(6)?

Investment implications

The range of productivity growth outcomes for the world economy in the years ahead has likely shifted higher. More growth basically, even if we should be very wary of those who would detail the geographical or even sectoral tilts of that potential uplift. Higher productivity growth could easily be linked to a different trend in real interest rates to that seen over the last couple of decades.(7) That alone is potentially extremely important for investors, as the influence of real interest rates is pervasive in investments.

This again reminds us that the decade ahead may look entirely unlike the ones in our rear-view mirror, with a very different investment winner loser board. Past performance really does contain very misleading information as the regulator rightly insists we remind investors. It is misleading because there is simply no requirement for the future to look anything like the past on which that investment performance must be based. What of the 10 years running up to 2020 told us anything about the years since for example?

Fortunately, this humility about the future is the guiding principle underpinning how we organise our multi-asset class funds and portfolios. The remarkable advances in AI these last few weeks and months represent an invitation to get invested and stay there. However, the more you focus your investments in an attempt to more surgically benefit from this new technology, the more you risk missing out on the real winners.

References

1 The Terminator (1984)

2 DeLong, Brad, Slouching to utopia (2022)

3 Briggs J & Kodnani D, The Potentially Large Effects of Artificial Intelligence on Economic Growth, 26 March 2023, Goldman Sachs

4 Briggs J & Kodnani D, The Potentially Large Effects of Artificial Intelligence on Economic Growth, 26 March 2023, Goldman Sachs

5 There is obviously no requirement for AGI to derive from LLM’s. As one critic has observed, LLM’s could easily be seen from the future as an off ramp on the route to AGI (Gary Marcus)

6 Bank Underground, Robot Macroeconomics: What can theory and several centuries of economic history teach us? September 2016

7 The London School of Economics and Political Science, Supply Matters, March 2023

Find out about our '?Ready-made investments ' via our Smart Investor platform. A selection of five Barclays funds that each aims to increase the value of your investments over time, using a broad mix of asset classes from across the globe.

Or

Learn about?Barclays Wealth Management , the affluent and high net worth service provider for Barclays UK.

*This article is for information purposes only. It is not intended as a product offer or investment advice