Let’s take a step back and think about AI…
Intro
These are times of transformational change, perhaps none more provocative than this disruptive new technology – AI – as it works its way into our economy, society, and lives. Much is being said about AI’s potential for outrageous and terrible impacts: from dumbing down of students, to widespread job destruction, to the extinction of the human race. Frightening stuff. With so much seemingly at stake, let’s take a breath and think.
Fundamentally, the widespread anxiety over AI lies in the very human tendency to fear the unknown. I will not try here to dismiss or refute fears. We simply don’t yet know what might happen. Many can quickly imagine unlikely worst-case scenarios; precious few seem able to envision a more positive future. Frequently cited concerns are legitimate and the outcomes are playing out in real time. Our response should not be panicked rejection; nor should it be apathy. It is a time to gain context and understanding. It’s about getting a workable and accurate perspective.
When considering the unknowable future, it can be instructive to think in terms of the bell curve of potential outcomes. What seems most likely and what seems remote? A highly improbable outcome that could have devastating negative consequences must be addressed, but more from a risk mitigation perspective than an all-in approach. There is no such thing as no risk and there are consequences to overreaction: an investor who goes all to cash may give up market returns – a mistake they might sorely regret at some date in the future.
Fear of the Unknown
Some might argue this point, but AI is not an inherently malevolent technology whose primary purpose is to destroy. That was the Manhattan Project. No, this is just a set of technologies –very powerful ones – that will have multitudes of uses, impacts, and outcomes. Most of these outcomes will in all likelihood be positive, generative, and productive for humanity. And there is the potential for harm. But how? Some basic questions seem to underly the broad fear of AI:
- Will it Hurt Us?
- Will it Diminish Us?
- Can we Trust the Founders?
- What Will Our Role Become?
These are big questions. Often it is the AI experts themselves who are the sounding alarm. If even the experts are scared, well then, we must really be…. But then. It is not at all the case that AI experts are unanimous in their opinions. Yes there are knowledgeable people making claims that should be taken seriously. But there are many others with similar intellectual authority who feel different: that the purported risks are overblown.
It is worth remembering that these experts may be engaged in some level of “self-dealing” or “talking their book” as some of the biggest alarmists about AI risks are also those who seek substantial funding to help them address said risks. And the regulations they support could prove anticompetitive in their favor. Maybe be skeptical of some of what you hear…
“Recursive Takeoff” is another source of heightened concern. This is the theory where AI does its own research and development: how might it evolve? There are many questions as to how far this can actually progress due to complications such as bottlenecks in computing and processing capacity, physical and hardware limits, diminishing returns on discovery of new algorithms and processes, and so on. In practice, this might prove much more theoretical than widespread reality.
AI is being used to address the problems it introduces. Much is being made about AI rogue agents and the risks if bad actors use AI for malign purposes. Often left out of this type of reporting is the far greater share of development targeted at safety and risk mitigation. For each threat that arises, a vast amount of industry and academic attention, money, engineering, tools, etc. come about in response. As challenges in containment, safety, and alignment occur, resolutions are discovered and implemented.
The argument against this is that we should be proactive rather than reactive in handling risks. The AI industry should take a safety mindset, like the airline industry, rather than react to problems as they come. This makes a lot of sense (thank you Nate Soares) because what if the harm was too great to respond to? Far better to move slowly and design/engineer safety into the process all along.
But remember that many people approach technological advancement with fear as they identify the jobs and skills that are likely to become obsolete or devalued. This glass-half-empty thinking ignores that throughout history, the net impact of new technologies has been more job creation than destruction. It’s just that people know the jobs that are being lost but are often unable to imagine yet the jobs being created. Recent articles in the MIT Technology Review and Barron’s each noted that the predicted job apocalypse appears to not yet have happened. Recent job numbers released by the Bureau of Labor Statistics support these observations.
Is it Humans Versus AI or Humans in Collaboration with AI?
Future AI capabilities may well be overstated. I’ve heard researchers and reporters gush about how AI superintelligence may soon outpace human intelligence, the implication being that AI will outpace us in all ways, like a new superior species, one that we have created. But this doesn’t have it quite right.
AI will of course be better than humans at tasks that require compilation and computation of massive amounts of data. Humans have never been that good at large scale rapid computation. We created mathematics, the abacus, calculator, desktop PC, and now AI, to aid us in our deficiencies. We see the need to use math and logic to solve problems and we use tools to help us crunch numbers. Bigger problems = more math = more tools.
But there are many areas where humans excel and AI is not capable. For example, anything where there is a need to understand one another, read each other’s emotions, or build trust; or where there is a requirement for wisdom, empathy, interpretation; or where there is an appreciation for human skill in creation or asking important questions. The list goes on and on. And compare the energy usage required to run AI processing versus the relative efficiency of human thinking.
What is exciting is the great potential that may arise when thoughtful and responsible humans work together with AI (and many other) technologies. This collaboration has the potential to bring about broad scale positive change for humanity. Are you someone who thinks that more of us have good rather than nefarious intentions? I hope so, because that is the case.
How Should Investors Think About This?
What to say about the assertion that AI stocks are currently in a bubble? Some have compared AI stocks to a bubble like the dot-com era of the late 1990s. This comparison is interesting and informative, but likely not yet accurate. The dot-com darlings had famously little (to negative) earnings; and the massive installation of “dark fiber” was telecommunication infrastructure for the new future that awaited. When the hype proved higher than the reality, the bubble burst.
The scenario today is quite different in that companies leading the AI transformation generate astounding levels of cash flow; and the tremendous expenditure into data centers is an effort to catch up with current demand not a promise of what demand could come in the future. Certainly, AI stocks could reach bubble territory, but it doesn’t seem we are there quite yet.
Also, there are claims that the economy and markets are being driven almost entirely by AI spending and that if this ends, there will be a resultant crash. The economy today has broader support. The large AI expenditures are certainly very positive for markets, but so is the highly resilient American consumer. Consumption spending has continued to hold strong, and accounts for between 65 – 70% of GDP. Finally, the banking infrastructure in the US is quite strong as well. Sure, there are headwinds that could become stronger, but the overall US economy and markets appear durable.
We are most likely still in an investing landscape where traditional approaches such as diversification and asset allocation should work well. We can’t yet know how the impacts from this transformative technological step will play out. If we had certainty, we could invest accordingly: concentrate into the most effective category. But since we don’t know this, we allocate across categories: some to handle potential downside risk, others to participate in positive responses.
Reminders of Positive AI Potential
For those who can see the potential for harm more than the possible benefits, let’s take a moment to recognize some of the areas where positive transformation may well occur:
Medical breakthroughs. Imagine personalized healthcare where diagnoses and therapies can be specifically tailored to the individual person, quickly and accurately. In this world there are portable MRI machines and optimized drug therapies. This could greatly reduce suffering and illness, improve outcomes and prolong lifespans.
Transportation. Transporting ourselves and the things we buy around the world is certainly one of the largest negative impacts on the environment (exhaust and waste) and quality of life (traffic jams). How we travel, commute, transport materials, etc. will most likely undergo dramatic change in the coming years and decades. I’d bet that a much quieter, cleaner, calmer, more efficient, and in fact better future awaits as these technologies continue to make their way into consumer and business acceptance.
Exploration. We barely know what lies at or even beneath the bottom of the ocean. We have so much to learn about the opportunities that will come from space travel. These aren’t just frivolous pursuits of billionaires. What if rare earth minerals were plentiful in in asteroids hurtling past earth? Also, what if key components to addressing climate change include technologies located in space? AI would be very important in design, testing, deployment, and maintenance of such tools.
Education. Each student, of whatever age, could have their own customized tutor that can work with students and implement approaches to teaching that are best suited to the individual student’s learning style. Wise savvy human teachers will employ these tools to radically change educational systems and outcomes.
Climate change. The increased demand for power imposed by AI technologies would be offset by myriad improvements in systems, processing of data, and planning for mitigation such as effective and appropriate use of geoengineering, CO2 removal, or space-based approaches. AI will likely improve the efficiency and reduce emissions for all manner of transportation technologies. It would help us make practical sense of complex systems such as weather, energy usage, traffic, migration, farming impacts, and industrial emissions.
Productivity gains. None of the above mention the increases in productivity that are highly likely to occur throughout all sectors across the economy. The current thinking is that we haven’t yet seen the gains companies are likely to achieve. As an investor, this is quite positive news as higher productivity can lead to higher valuations.
Conclusion
Anxiety about AI cuts across the spectrum of affiliations, viewpoints, status, and expectations. A review of headlines during other historical moments when society was undergoing great technological change shows how the language was quite similar to what we’re seeing today. Many were quick to imagine devastation to livelihoods, job categories, society, even the human race as we know it. The steam engine, electricity, rail roads, the assembly line, telephones, desktop computers, space travel, the internet, the smartphone, all inspired fearful responses that now look like overreactions given our perfect 20/20 hindsight. I wonder what were early reactions to fire and the wheel.
Yes, we should carefully think through the potential harms that might come. But we’d be foolish, alarmist, to live only in fearful narratives without considering context. We should give at least as much attention to the vast opportunity for enduring positive results: many of which we already know of, but many more that we can’t yet even imagine. If AI helps us solve the complex questions we now face, then we would have more time to create and address all new questions.
Securities and advisory services offered through LPL Financial, a registered investment advisor. Member FINRA/SIPC.
This material is for general information only and is not intended to provide specific advice or recommendations for any individual. There is no assurance that the views or strategies discussed are suitable for all investors or will yield positive outcomes. Investing involves risks including possible loss of principal. There is no guarantee that a diversified portfolio will enhance overall returns or outperform a non-diversified portfolio. Diversification does not protect against market risk. The economic forecasts set forth in this material may not develop as predicted and there can be no guarantee that strategies promoted will be successful.
The Consumer Price Indexes (CPI) program produces monthly data on changes in the prices paid by urban consumers for a representative basket of goods and services (Source: U.S. Department of Labor).
The PCE price index, or Personal Consumption Expenditures price index, is a measure of the prices of goods and services purchased by people in the United States.