The era of personalised AI agents may be closer than imagined even a couple of years ago. Software-based assistants doing a variety of tasks, from booking flights to ordering pizza, will create the mechanised consumer, who will then force producers to build their own mechanised interfaces to operate in an automated marketplace. Personal agents would, indeed, be a breakthrough for a technology that's yet to find its feet among enterprise users. Yet, there may be some heroic assumptions involved in the notion, chiefly surrounding the tricky issue of trust.
At one level, personal agents will have to overcome consumer concerns over privacy. To be able to organise a holiday, an AI agent will need access to an individual's emails, text messages, browsing history, bank account and credit cards. Assuming all these have been granted, the agent will have to interface with airlines, hotels and visa services. Will they be comfortable transacting with an agent on an individual's behalf? They should, if enough agents were bringing them business. But then they, in turn, would have to automate their interface for the new customer avatar. And all transactions between the buyer's and seller's agents would have to be secured.
If this sounds like science fiction, consider what it means for industries like advertising. There wouldn't be much point selling one beach over the other - the AI agent would know. A marketplace where AI agents transact with each other would have no information asymmetry, an impossible condition in markets where humans operate. There would be no need for regulators to uphold consumer or producer rights. The role of governments would shrink (alas, not disappear). Yet, we are closer to this utopia than we think - at least according to Meta, which has almost half the world's population interacting on its platforms. Meta runs on advertising and the bots it is working on are designed to operate in an ad-free environment. It will be interesting to see how it resolves this conflict.
At one level, personal agents will have to overcome consumer concerns over privacy. To be able to organise a holiday, an AI agent will need access to an individual's emails, text messages, browsing history, bank account and credit cards. Assuming all these have been granted, the agent will have to interface with airlines, hotels and visa services. Will they be comfortable transacting with an agent on an individual's behalf? They should, if enough agents were bringing them business. But then they, in turn, would have to automate their interface for the new customer avatar. And all transactions between the buyer's and seller's agents would have to be secured.
If this sounds like science fiction, consider what it means for industries like advertising. There wouldn't be much point selling one beach over the other - the AI agent would know. A marketplace where AI agents transact with each other would have no information asymmetry, an impossible condition in markets where humans operate. There would be no need for regulators to uphold consumer or producer rights. The role of governments would shrink (alas, not disappear). Yet, we are closer to this utopia than we think - at least according to Meta, which has almost half the world's population interacting on its platforms. Meta runs on advertising and the bots it is working on are designed to operate in an ad-free environment. It will be interesting to see how it resolves this conflict.
