Making predictions is always hard, especially in an industry moving as quickly as artificial intelligence. But as we stand on the cusp of a new year, it’s worth reflecting on the progress AI has made over the past twelve months, as well as what might lie ahead for the technology in 2024.
2023 was undoubtedly a monumental year for AI, but fear and speculation often dominated the headlines. 2023 saw the rise of the AI doomers, with online commentators like Eliezer Yudkowsky rising to prominence for warning that AI poses existential risks to humanity. Adding to this air of pessimism was a high-profile open letter from leaders of the tech industry, calling for a 6-month pause on the most powerful AI research so experts can assess risks and create safety protocols.
Whatever the merits of such proposals, slamming the brakes on AI is unlikely to happen any time soon. Even if it were politically feasible to stop progress, it would be a strategic blunder to do so, when this would only put western democracies at a disadvantage relative to adversaries like Russia and China.
Furthermore, even those actively calling for a pause don’t seem to take the idea seriously. Signees of the letter, like Elon Musk, are moving forward with their own AI models regardless. Grok, the AI chatbot developed by Musk’s company, xAI, is the latest example.
Much of the climate surrounding doomerism has been fueled by the “effective altruism” (EA) movement, which gained significant public attention in 2023. Yet, despite the movement’s deep pockets and rising influence in the corridors of power, EA has had a tough couple of years. The inept firing and rehiring of OpenAI CEO Sam Altman by the company’s EA-affiliated board, along with EA’s close connections to cryptocurrency fraudster Sam Bankman-Fried, have shown the movement is not always as effective or altruistic as advertised.
That said, doomerism doesn’t stand or fall based on its connections to a dubious pseudophilosophy. Undoubtedly, many AI risks are real and should be taken seriously regardless. The national security implications of AI in particular constitute perhaps the biggest area of concern. AI will enable new forms of cyber warfare targeting critical infrastructure. It could facilitate dangerous biological and chemical weapons development. And it will propel information warfare to new frontiers through advanced propaganda and deepfakes.
In other words, issues that were once Hollywood fantasy are fast becoming reality, as biometric technologies start to permeate daily life. Facial recognition is already becoming widespread at airports. AI is transforming law enforcement, perhaps heralding the arrival of a mass surveillance state. Netflix’s
NFLX
One key battleground in the year ahead will be between open and closed AI systems. Ironically, OpenAI and spinoff Anthropic now represent closed black-box models, while companies like Meta embrace openness. Regulation will heavily influence whether opaque systems, like ChatGPT, or transparent one’s, like Meta’s LLaMa, ultimately prevail.
The European Union’s pending AI Act lays bare the tradeoffs policymakers face balancing oversight with innovation. Excessive regulation risks entrenching Silicon Valley tech behemoths while hurting startups and open source competitors. If Europe has any hope of innovating in AI itself, it is going to have to foster a competitive marketplace that enables upstarts like France’s Mistral AI to thrive. Instead, the EU may legislate these promising companies out of existence.
Back in the US, President Biden’s recent executive order on AI has mostly delayed specifics, but the regulations it mandates are coming down the pike in 2024 and will certainly shake up America’s AI policy landscape. Debates around some issues, like bias and discrimination, will likely fall along predictable partisan lines. But knotty problems surrounding copyright and IP are more complicated.
In what could be a groundbreaking case, the New York Times
NYT
Another dispute will arise surrounding the industry’s increasing energy usage, given AI models need massive datacenters to power them and these use a lot of electricity. The debate over AI energy use is likely to closely mirror fights over the environmental toll of cryptocurrency mining. The issue will only grow more divisive as the competition over “compute” reaches new heights.
In spite of these controversies, 2024 will hopefully be the year rationality and nuance retake center stage in AI policy. America and Europe can still craft prudent governance strategies that protect freedom while securing the benefits of AI. But to do so, our leaders need to anchor discussions in facts and evidence, not unfounded emotions.
Both blind optimism and crippling pessimism impede thoughtful solutions. The stakes are immense, but the possibilities are also astounding. 2024 promises to be a pivotal year as society grapples with AI’s double-edged sword.