Unlock the Editor’s Digest for free

Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.

This has been a crazy, breakout year for artificial intelligence, which became the hot topic for discussion in almost every corporate boardroom. More than 100mn users experimented with OpenAI’s ChatGPT chatbot and other text and image generation services, churning out countless school essays, personalised marketing pitches and viral fake images, including the Pope in a puffer jacket.

The British Prime Minister Rishi Sunak took a break from grappling with hospital waiting times and migrant boats to host a two-day conference at Bletchley Park to explore the technology’s existential risks. And the San Francisco start-up OpenAI succumbed to a bitter, and somewhat bizarre, boardroom row in which chief executive Sam Altman was fired and then rehired a few days later after a staff revolt.

Yet, in spite of all the hype and the drama, few compelling business use cases for the technology emerged. All the talk about generative AI reinventing internet search fizzled out after Microsoft’s AI-enabled Bing failed to disturb Google’s market dominance. Concerns about data security, intellectual property rights and generative AI’s dirty habit of “hallucinating” facts — or, more crudely, just making stuff up — also deterred companies from deploying the technology. And many AI-powered start-ups, promising to revolutionise various industries, blew up on the launch pad as the release of increasingly capable generative AI models destroyed their original business models.

All this is likely to change next year as AI moves from fun experimentation to focused adoption. The big technology companies are re-engineering their entire organisations around AI, embedding it in almost all their services. Many other companies have also wised up to the technology’s possibilities, and limitations, and will use it to sharpen their own competitive edge. Some creative start-ups will undoubtedly find valuable use cases for generative AI in narrow, specific domains.

This evolution will be accelerated by three trends. Some generative AI models are already becoming smaller and more accessible and will soon become usable on smartphones, as Apple is now planning. The proliferation of open-source models will help companies to deploy generative AI safely on their own proprietary data sets for clear business use cases. The launch of more powerful multi-modal models, blending text, image, audio and video, will also extend the creative possibilities. 

The tech companies say they can deal with many of generative AI’s flaws themselves. Some AI company executives accept the principle that they should pay something for the swaths of intellectual content they suck in to train their models. This month, OpenAI struck a landmark deal with Axel Springer to pay the German publisher for access to its journalism. Tech companies also promise to introduce watermarking of synthetic content to help prevent the spread of disinformation.

Still, governance will remain a big issue next year. The leading AI companies, including most notably OpenAI, have some way to go to convince outsiders that they can be trusted to manage such a powerful general purpose technology. The US administration is already adopting a more activist approach, having signed an executive order instructing many federal agencies to regulate AI’s uses. The EU has this month passed an AI Act limiting the technology’s use in high-risk areas. The Chinese government has also adopted highly stringent rules to control the technology.

Our collective obsession with AI is only likely to intensify next year as — in the words of the science fiction writer William Gibson — the “street finds its own uses for things”.