The page you were trying to view is not available for your role.
Summary
The debate around AI is shifting from whether it will transform the economy to how it should be developed and governed. While AI has the potential to boost productivity, improve decision-making and drive company profits, concerns remain around security, regulation and the wider impact on jobs and society. The latest blog from Portfolio Manager, Ian Jensen-Humphreys, argues that AI's long-term benefits are significant, but that governments, regulators and technology firms need to work together to ensure it develops in a safe, responsible and sustainable way.
"The future is not set. There is no fate, but what we make for ourselves."
Thirty-five years after Terminator 2, that line feels surprisingly relevant to the AI debate. Anthropic CEO Dario Amodei has recently been publicly calling for caution in the build-out of AI and to be aware of the inherent risks. Unsurprisingly, since his company Anthropic (the maker of Claude) is one of the largest AI hyperscalers, he is not calling for AI development to stop. Rather, he is arguing that the industry should proceed more deliberately, giving safety, oversight and understanding a chance to keep pace with capability. The question for investors is not whether AI will transform the economy, but whether the path to that future needs fewer accelerators and more brakes.
"Skynet became self-aware..."
So, what are the potential issues? At the risk of introducing some hyperbole, the risk is that we (or rather one or more of the hyperscalers) develop an AI model or agent that becomes sufficiently sentient that we (the human developers) lose control of it, with devastating consequences. Setting aside the “AI will kill off the human race” fears, there are more mundane but real-world issues.
A good example of this would be Mythos, a model developed by Anthropic to test the Pentagon systems by analysing their cyber security processes and finding and reporting flaws in the software. It appears (although I’m sure we, the public, are only receiving the sanitised story) that the model was able to break into nearly all of the classified systems in hours, rather than weeks as had been expected. Clearly if this were to fall into the wrong hands, the consequences could be very serious. Alternatively, what if Mythos simply “decided” to release all the hacked data into the public domain without human consent or direction?
"That's a nice bike."
How do we fix these potential issues? Likely with some form of regulation – either self-regulation by industry groups or alternatively, more onerously from governments. An apt analogy would be regulations around cars or motorbikes – the benefits are clearly transformational in terms of speed of travel, but the danger from accidents is real and present. So, we have regulations to mitigate the danger – speed limits, tests to obtain driving licences, minimum vehicle maintenance standards etc. It hasn’t stopped cars and bikes from becoming ever more efficient and useful over the years.
The harder question though, is how might a government regulate AI? Can it even be done at a country level? Or might there need to be a globally agreed framework? If so, countries might differ in their desire to regulate depending on whether they have a large domestic AI industry – impacts on local jobs can be a persuasive factor.
"This is the world now. Logged on, plugged in, all the time."
So why bother at all with AI if it costs so much money to build out and comes with moral hazard risk? From the perspective of companies, AI has the potential to materially increase their productivity and generate valuable insights from the data that they collect through their operation. Examples of this could be insurance companies developing better underwriting risk models, or biotech companies conducting virtual product development. At a more day-to-day level, AI agents can build presentations, write computer code or draft legal documents far more efficiently than any human can. All of this could translate into greater profits and hence share price gains. We’re already starting to see some companies reference AI related gains as they announce their latest corporate earnings, and this is only going to continue.
But is this definitely a good thing? More work being done by AI agents means potentially less work available for humans to perform, with potential knock-on impacts for wages (lower) and unemployment (higher). Will most of the benefits of AI accrue to companies rather than individuals? For government and central banks, the question is whether AI will help deliver the “right sort of growth” and how can they help those cohorts of people who lose out as a result of AI cope with a structural change in their circumstances.
"Judgment Day is inevitable."
Whatever the latest headlines, the reality is that AI is not going away any time soon. It has the potential to have a transformational effect on many aspects of our lives – either by increasing our productivity at work, saving us from having to do mundane, repetitive tasks or helping speed up medical advances.
Like any new “invention”, AI can be used for both good and bad. It is up to governments and companies to agree on frameworks and boundaries to help as many of us as possible benefit in a safe and sustainable way.
Key takeaways
- AI offers major growth potential - AI is already helping businesses improve efficiency, analyse data more effectively and reduce costs, which could support future earnings growth and investment opportunities.
- Regulation will be crucial - As AI capabilities advance, the focus is increasingly turning to governance and oversight. The challenge will be creating frameworks that encourage innovation while managing risks.
- Winners and losers are likely - While AI could boost economic productivity and accelerate innovation, it may also disrupt labour markets and change the way people work, creating both opportunities and challenges for policymakers and businesses.
Ian Jensen-Humphreys
Portfolio ManagerIan is a portfolio manager of the Quilter Investors Cirilium and Creation Portfolios. Ian joined Quilter Investors in March 2020 from Seven Investment Management (7IM), where he was deputy chief investment officer. Ian also spent 15 years at Goldman Sachs in risk management and portfolio hedging strategies.
Ian is a CFA charterholder and has a degree in Physics from the University of Oxford.
