THINKING ALOUD

When AI agents haggle, who gets stuck with the bill?

The technology promising to save us money could escalate a costly arms race between shoppers and businesses

Summarise
    • US retailers use AI for “surveillance pricing” to find out which consumers will go or stay at given price points.
    • US retailers use AI for “surveillance pricing” to find out which consumers will go or stay at given price points. PHOTO: PIXABAY
    Joyce Hooi
    Published Tue, Nov 10, 2026 · 07:00 AM

    MY BURGEONING partnership with my artificial intelligence agent, Meta’s newly launched Muse, has hit a snag.

    The eager assistant had found me a plan from my existing Internet service provider (ISP) that is similar to my current one, but cheaper.

    So, it asked me if I wanted it to call my ISP’s customer service hotline and “ask for the retention team”, with the aim of lowering my Internet bill.

    “Are you gonna talk to them for me, or do I have to speak to customer service myself?” I’d asked Muse. Here was where it went pear-shaped.

    “Honestly, you’ll need to speak to them yourself – I can’t make a live phone call (to the ISP) for you,” Muse told me.

    Well, nuts to that. I did not painstakingly customise Muse’s avatar as a chubby British shorthair cat and dress it in a little pink sweater, only to be told to speak to another human being on the phone myself. This is 2026, good grief.

    Yet for every phone-allergic millennial like me, there must be multiple extraverted and driven people who would have jumped on the phone to get a better deal.

    The idea here is that the consumer AI assistants will find us the best offerings and nudge us into overcoming our pricing and subscription inertia, saving some money in the process.

    But mind you, Muse and its ilk are merely the latest foot soldiers in a pricing war waged quietly on consumers for years.

    Pricing dynamics

    In the US, retailers have been using AI to turbocharge “surveillance pricing”, running large pricing experiments to assess a consumer’s willingness to pay.

    Thanks to AI pricing software, some American grocery shoppers are unwittingly paying up to 23 per cent more for the same thing than their peers are.

    Now, agents like Muse are ironically supposed to help us right this wrong, enabled by the very technology underpinning it.

    With AI on consumers’ side, prices should ideally converge downwards to what the market will bear, under the unrelenting pressure of agentic interrogation.

    I reckon, though, that the escalation will simply worsen. The agentic layer in corporations will thicken in response to consumer bots, as firms devise more elaborate ways to distinguish customers who will leave from those who will stay.

    So even as consumer agents work hard to secure concessions, corporate systems will work harder to ration them.

    This escalation will ultimately drive up costs that will likely be passed on to consumers. Shareholders rarely get stuck with the bill; S&P 500 firms collectively booked a 30.2 per cent growth in earnings year on year in the third quarter, even as workers’ real wages fell.

    There is also the danger that an individual convenience will become a collective bare minimum. Once everyone has a tireless AI assistant, lacking one may mean accepting worse terms as a consumer.

    Eventually, AI agents risk trapping us in a loop of busywork that might lead back to prices inflated by the cost of all this agentic activity.

    The winners in this arms race might then be the hyperscalers supplying the token ammunition, like Meta.

    I hope I am wrong about this, but I am hedging my bets by having a Muse agent on my phone and Meta shares in my brokerage account.