Riley Lankes
21st February 2024
*This briefing was written with the help of generative AI. For more details, see SLAAIT’s AI training resources on our website*
You may have heard talk that we’re experiencing an AI arms race right now, but that isn’t quite accurate. We’re actually experiencing two.
Artificial intelligence is quite an expensive technology to develop, especially if your goal is to have control over what you create. To clarify here, we’re talking about large-scale AI models such as ChatGPT, or Google’s newly renamed Gemini. Even if you can fund the small army of talented programmers and AI experts you’ll need to develop and maintain your AI model, you’ll still need hardware to run the model on. The computing hardware needed to run a large model is incredibly expensive, putting the ability to develop and run these types of AI models out of reach of all but the most financially robust organizations. It should come as no surprise then that the arena is largely dominated by governments and tech giants.
Governments and tech giants have very different motivations for developing large-scale AI, and it’s these differing motivations that have resulted in the two AI arms races we’re witnessing right now.
Governments Race to Develop AI
Governments, perceiving myriad military and civil applications for AI, are driven to research and develop the technology. However, in an effort to keep any discoveries or developments out of the hands of their adversaries, state-backed research into AI is largely kept secret. Even so, given what information has been made public, we know that countries like the United States, China, and Russia are actively developing military applications for AI.1 We’re also already witnessing countries taking measures to prevent their geopolitical adversaries from accessing AI technology, as seen in the US restriction of Nvidia AI chip sales to China.2
Tech Giants Race to Deploy AI
In contrast to the largely obscure race between international governments, the second arms race unfolds in a markedly different arena – the competitive landscape of global tech corporations. Here, the race is not solely about technological innovation but also about swift deployment – it’s a race to capture the market. The rationale is clear: the early capture of a vast user base can catapult a company to market dominance, as exemplified by the meteoric rise of social media giants like Facebook (now Meta) in the early 2010s. Companies are fervently offering free AI services to users, alongside rapidly integrating AI functionalities into existing software.
This frenzied pursuit of market share comes with inherent risks. The business model adopted by tech firms prioritizes rapid development and deployment of AI technologies, often at the expense of thorough testing and risk assessment. Interviews that I conducted with industry insiders in 2023 corroborate this sentiment. The mantra becomes “put it out there and see what happens,” with end users becoming unwilling beta testers for the tech companies providing these free AI services.3 This results in the widespread deployment of AI to millions of users with minimal pre-release scrutiny.
This hasty deployment of AI technology has profound implications, potentially exacerbating existing societal issues. From the misuse of generative AI for disinformation campaigns to the erosion of trust in political institutions and the veracity of video content, the repercussions are far-reaching and multifaceted. My own research uncovered evidence to suggest that this corrosion of political trust is already well underway.4 Another study focused on propaganda recently found that AI-generated propaganda is nearly as persuasive as materials put out by from real-world foreign covert propaganda campaigns, while being significantly easier to create. 5Even so, in the rush to capture market share and reap financial rewards, these concerns are seemingly waved away.
Critical Takeaways
The two AI arms races unfolding before us in 2024 – one between governments, the other between tech giants – underscore the complex interplay between technological advancement, market competition, and societal impact. As the trajectory of AI development accelerates, it becomes imperative to strike a balance between innovation and responsible deployment, lest the pursuit of profit outpaces the safeguarding of societal well-being. Only through careful consideration and ethical stewardship can we navigate the evolving landscape of AI with prudence and foresight.
- See DARPA’s AI Next Campaign, DoD reports on China’s Military AI, and analysis of Russia’s use of AI ↩︎
- Full story from Reuters ↩︎
- Read my full paper on this topic, available at corrosiveai.com ↩︎
- See Corrosive AI ↩︎
- How persuasive is AI-generated propaganda? ↩︎