| Authors around the world are weighing how to respond to the emergence of generative AI. Some are demanding assurances that their work will be excluded from AI training data. Others are reluctantly striking deals to get paid for their work’s inclusion, as Alice Robb writes in an essay this weekend. Microsoft offered publisher HarperCollins $5,000 in exchange for allowing the tech giant to use Robb’s 2018 book in its training data. Her decision (she’d get half the cash) was clouded by uncertainty: No one could say whether this was a good deal. “Anything’s better than zero,” the financial journalist Felix Salmon told her. Another writer friend argued for taking it, citing “desperate times.” But while much of the discussion so far has been about getting work out of AI, some writers are wondering how they can get their work into AI — in the hopes of influencing what it has to say. Take a post on LinkedIn by Ken Mandel, a marketing executive at transport and payments app Grab, that made the rounds last week: “For over a century, advertising has been about human attention,” he wrote. But that changes if AIs, not humans, make purchasing decisions. If an AI agent is the one picking out your groceries for you, who does a marketer need to convince? “In a world where agents (not humans) interact with brands, the game shifts from advertising to AI influence and optimisation,” Mandel mused. “Brands must ‘train’ AI agents to prefer their products in natural decision-making.” Economist and Bloomberg Opinion columnist Tyler Cowen recently made a similar argument, but for public intellectuals. “Consider the AIs as part of your audience,” he advised. Write to teach them things, to expand your influence and to persuade them of your point of view. That idea got a boost this week with the launch of OpenAI’s “deep research” tool, which Cowen compares to “having a good PhD-level research assistant” — except much, much faster. If AI agents increasingly do research for journalists, politicians and businesses, then academics, policy wonks and activists may join marketers in the race to influence what those agents think and say. And if marketers and public intellectuals are actively trying to get AI to pay attention to their work, that means more content creators choosing to let AI train on their data for free — which could make it harder for other authors to get paid for granting the same access. In the end, Robb took the deal from Microsoft. The horse was out of the barn, she concluded. — Walter Frick, Bloomberg Weekend Edition | 