SKY Eye

The much-hyped release of a new OpenAI model has potentially huge ramifications and could completely change the way we interact with AI.

When OpenAI released ChatGPT in November 2022, it proved to be a seminal moment. It kick-started the generative AI revolution, which I’ve written about in my new book, How AI Ate the World. But a new release from the same company may prove even more impactful.

On May 13th, OpenAI announced the release of GPT-4o, a multimodal version of GPT-4 with plenty of improvements that looks set to change the way people interact with generative AI.

The “o” in GPT-4o stands for “omni,” OpenAI said, and the company sought to demonstrate how the AI model could be used in a number of different ways and formats through its tech demo, which was livestreamed on the internet.

OpenAI demonstrated GPT-4o’s dexterity in handling live video, audio and text – something that made it far more human-like than previous generations of the ChatGPT chatbot, which mainly used text to interact with its users.

Science fiction becomes science fact

The smoothness with which the demonstrators could probe and chat with the AI model in real time could prove a game-changer for its adoption, especially by those who have so far shied away from trying ChatGPT or other generative AI chatbots. Some users haven’t bothered trying out the latest technology because they think the text chat interface is clunky, and it reminds them of previous unhappy interactions with less intelligent chatbots.

For that reason, GPT-4o could be the thing that takes generative AI from the precipice of world domination to outright ubiquity. Already, companies have recognised the impressive power of the generative AI revolution – and the risk of being left behind. Goldman Sachs estimates that the tech could help boost global GDP by $7 trillion in the next decade. The International Monetary Fund reckons that 40% of all jobs could be affected in some way by generative AI.

The science fiction-style future we’ve been thinking of and writing about for decades seems to be here. A chatbot that we can converse with, that can recognize our emotions and respond accordingly sounds like a dream. For those who had yet to be convinced by the power of ChatGPT and refused to get caught up in the hype, GPT-4o may make adoption easier. When a technology seems eerily human-like, it’s very hard to refuse to use it.

A second hype cycle

The irony is that GPT-4o arrives at a time when the initial AI hype cycle is starting to slow. Eighteen months after the release of ChatGPT, people were beginning to question whether the tech and its adoption had plateaued. Around the same time as OpenAI was releasing information about its model, reports surfaced that mentions of “artificial intelligence” or AI in quarterly earnings reports from companies had nosedived.

That’s important because despite the ubiquity of the technology in our world and our businesses, most Americans, for instance, haven’t used tools like ChatGPT, according to Pew Research Center.

Fewer people say they use ChatGPT for entertainment or to learn something new than in the course of their work, suggesting that many of those who are using the tool are doing so because they see an economic benefit to it, rather than necessarily out of choice.

But the new format of interaction and the increased smoothness with which people can use GPT-4o may start to see that number increase. The reduction in friction could well be a boon for those who want to adopt ChatGPT – with potentially transformative effects on our lives.