Will Generative AI do more harm than good?

Marcus on AI | Substack

Source published: Site published:

Original article · فارسی

Things aren’t looking good.

In early March 2023, I debated Keith Teare in an event hosted by the organization Open to Debate. We discussed whether ChatGPT would do more harm than good. I think about this question a lot, almost every day.

Right now, the arguments against generative AI—the list of bad things it can do—are very serious. These include fake sexual images and videos made without people's permission, more scams that trick people into giving away private information, and large amounts of false information. Another concern is that people may let AI do their thinking for them instead of thinking for themselves. Governments could also quietly influence large language models (LLMs), which could threaten democracy. Generative AI can also be used to watch and track people. On top of this, AI systems have used artists' and writers' work without permission. AI could also hurt jobs across the economy, although thankfully, the effects on jobs have been smaller so far than some people expected.

There is also the huge wave of computer attacks that some people have called “Black October”.

Paul Kedrosky has called this wave of attacks The Hackening.

Marc Benioff, the CEO of Salesforce, called the situation a “Code red.” He shared a worrying post by Anthropic's Jason Clinton. I think both of them are right. However, Clinton believes this very difficult period might last only a year or two.

X avatar for @Benioff
Marc Benioff@Benioff
Jason is completely right. Code red. 🚨
X avatar for @JasonDClinton
Jason D. Clinton 🔸 @JasonDClinton
Things have become very bad over the past 4 days, and they will get much worse before they get better. Security leaders, use this crisis as a reason to act. Please ask your company boards for help right now—not to fund AI, but to cover the basics. We need to get through about 18 months.
11:09 PM · Oct 22, 2026 · 844K Views

89 Replies · 159 Reposts · 2.01K Likes

But what about the benefits? Are they worth all these risks? Torsten Slok, the chief economist at Apollo Global Management, has sometimes been quite positive about this technology. But he recently made a strong argument that many of the benefits we were led to expect simply are not appearing:

You can read the full article here.

This story is not over yet. I know people who have gained a lot from generative AI, mainly by using it to write code. But right now, it is not hard to argue that, overall, generative AI has done more harm than good for society.

There is also still a chance that it could bring down the entire economy.

Get ready! There may be a rough road ahead.

Bonus: I will return to Open to Debate on Tuesday in Baltimore, where I will take part in an event with Cathy O’Neil. There will be no public livestream, but edited versions will eventually be available on YouTube and through many NPR stations, including WNYC.