AI Is Not the New Crypto

[Celerity]

The torrent of investor money that flowed into crypto is now hitting the AI scene. We’re already seeing the results.

https://www.theatlantic.com/newsletters/archive/2023/01/ai-is-not-the-new-crypto/672746/

https://archive.ph/WtBsR

Recent breakthroughs in generative AI, such as the image generator DALL-E and the large language model ChatGPT, are “potentially akin to the release of the iPhone in 2007, or to the invention of the desktop computer,” Derek Thompson told me in December. Here are the latest AI developments to watch in the coming weeks and months.

Hype Machines

Investors are pouring money into AI.

Last year, investors put at least $1.37 billion into generative-AI companies across 78 deals—almost as much as they invested in the previous five years combined, according to the market-data company Pitchbook. Microsoft, in particular, has taken a big leap: Since 2019, the company has invested $3 billion in OpenAI, which designed DALL-E and ChatGPT, and it’s reportedly in talks to invest another $10 billion. Microsoft purchased an exclusive license to some of OpenAI’s technology, and it’s working with OpenAI on a new version of its search engine, Bing, that would incorporate a ChatGPT-like tool.

Schools are concerned about academic integrity.

How will these tools change our lives? As Derek told me recently: “We don’t know. The architects of those technologies barely know. But it’s so interesting to play with, and the technology is improving so quickly, that we should absolutely take it seriously, as if it’s something that can’t be avoided.” Some universities are modifying their courses to minimize the risk of students handing in essays generated by an AI tool. And they’ll likely have to deal with even more capable tools soon—OpenAI reportedly plans to release GPT-4, which would be better than the current versions at generating text. Meanwhile, a 22-year-old computer-science student has built an app to identify whether a piece of text was written by a bot.

It may be time to worry about deepfakes—again.

You might remember that term from back in 2018, when media outlets and misinformation experts panicked about a rise of fake, realistic-looking videos. (In a famous example that BuzzFeed engineered, Barack Obama appeared to say “President Trump is a total and complete dipshit.”) While that panic remained just that—a panic—advances in generative AI “have experts concerned that a deepfake apocalypse” is on the horizon, our assistant editor Matteo Wong reported last month. As AI-generated media get more advanced, these experts argue, in the next few years the internet will be flooded with forged videos and audio touting false information.

snip