America is facing a crisis of public trust, whether it’s the government, the media, the Federal Reserve, or simply those with opposing political views. The artificial intelligence revolution is happening in America as AI transforms everything from Google’s business models to our childhoods.
Looking at these two developments made me think: Could the AI revolution somehow be harnessed to bring about a restoration of social trust, rather than a further breakdown of it?
I admit it’s a tall order. Social media, the latest technology to take over the Internet, is often associated with an increase in misinformation and, in turn, a decline in social trust. And much of the recent commentary on AI points to these same risks. Using Large Language Models (LLM) can create an enormous amount of propaganda and confuse the Internet.
It’s a real risk, but it doesn’t seem all that different from the status quo. Tyrants and villains are already hiring humans to flood the Internet with malicious content.
The more hopeful news isn’t about content, it’s about curation. Current leading LLMs such as Anthropic and OpenAI, which are trained using internet data, provide relatively objective answers when asked about Russia or China. Users can also create a more factual or academic answer by requesting it.
The point is not that LLM will not be used to create propaganda, it will be used to create propaganda, but that it will give users another option to filter propaganda. is. LLM gives users the desired degree of objectivity, at least once they learn how it works.
Still, this is a safe prediction. Within a year or two, there will be various LLMs, including open source ones, that people will be able to use to generate the kind of answers they want. Not to mention all sorts of voices on Twitter, viewers may wonder how this will improve the current situation where he now has a choice of polarizing media sources such as Fox News and MSNBC. not.
Partly because LLM works on a “pull” basis, meaning you ask LLM for what you want, I have hopes for improvement. Even when working with a “right-wing” LLM, you can always seek left-wing or centrist perspectives, and vice versa. It’s like watching Fox and having a button on your remote that, when clicked, gives you an opposite or contrasting view within seconds. This is a huge improvement over cable TV in terms of immediacy above all else. LLM also makes it very easy to generate discussion and “contrast” answers for almost any issue.
Again, I don’t know how much balance people want. But at least the AI will make it easier for them to balance if they want to. This seems better than using a specific cable TV channel or a pre-established Twitter feed as a baseline default.
The impersonal nature of many LLMs can also be an ideological balancing force. Now many leftists are unwilling to switch over to Fox (or visit their neighbors) to hear another perspective because their personalities are aggressive or obnoxious. LLM offers the possibility to sample perspectives in the least insipid and provocative, and most analytically debated form.
LLMs can also increase trust by making translations between languages nearly seamless. Some opponents of immigration would trust Latino immigrants more if they could sample Spanish-language news and media to better understand the very real problems immigrants face in their home countries. Are you there? Call me naive, but I think there is more potential for upside than downside. People who are more exposed to different cultures tend to be more tolerant and more trusting.
LLMs can also increase credibility in areas beyond politics. For example, doctor visits can be more meaningful and productive when an LLM helps you prepare the right questions about a health issue. You and your doctor may be able to build a better rapport.
America’s crisis of social trust has complex causes, not all of which are known, and AI is by no means a panacea. But at least LLM has the potential to remove the current bias towards balance and make a wider view more widely available.
Of course, we humans have to decide how much we really want to trust each other. If we can’t do this right, even the smartest AI implementation will be useless.
Tyler Cowen is a Bloomberg Opinion columnist. George is a professor of economics at Mason University and writes on his blog, Marginal Revolution. He is co-author of Talent: How to Identify Energizers, Creatives, and Winners Around the World.
