Overview of Bloomberg’s Odd Lots Episode: AI Agents Are Already Changing the Price of Goods
This episode explores how AI agents and algorithmic pricing are reshaping shopping, pricing, and competition across the economy. Hosts Tracy Alloway and Joe Weisenthal speak with Lindsay Owens, author of Gouged: The End of a Fair Price and What That Means for Your Wallet, about surveillance pricing, personalized offers, and the emerging “arms race” between AI tools used by consumers and by sellers. The core theme: pricing is becoming less transparent, more individualized, and increasingly driven by data, experimentation, and AI.
Key Themes and Main Takeaways
1. Pricing is no longer simple sticker pricing
Owens argues that companies have moved far beyond the old model of “cost plus margin.” Today, pricing is often:
- algorithmic
- dynamic
- personalized/surveillance-based
- optimized through constant experimentation
She says companies now use AI, data, and behavioral signals to determine “the optimal price” for each consumer.
2. AI shopping agents can help consumers — but may also work against them
The hosts discuss AI assistants like Muse as potential “personal shoppers” that could:
- compare prices
- cancel subscriptions
- save consumers time and money
But Owens warns that agents may not always act in the user’s best interest. In tests she cites, agents sometimes:
- found more expensive options even when asked to find the cheapest
- surfaced sponsored products instead of the lowest-priced ones
3. Sellers have AI too, creating an arms race
A major point of the conversation is that buyers may get AI tools, but sellers will too. That means:
- retailers can use much richer data to infer willingness to pay
- algorithms can test different prices in real time
- consumers may end up in an escalating battle against systems designed to maximize revenue
Joe Weisenthal frames this as a potentially wasteful “arms race” where both sides pay AI companies, yet prices may end up roughly where they would have been anyway.
4. Fair pricing is becoming less common
Owens argues that fixed public price tags were historically important because they:
- created predictability
- enabled comparison shopping
- kept prices competitive
- protected budget-conscious shoppers
She says modern pricing systems undermine those benefits by making prices variable and opaque.
Evidence and Examples Discussed
Instacart pricing experiment
Owens describes a study where 400 secret shoppers used Instacart simultaneously in different places to buy the same groceries from the same store. Findings included:
- 75% of items had different prices across shoppers
- price differences reached as high as 23%
- some families could pay up to $1,200 more per year
This was presented as evidence that personalized pricing is already widespread.
Expedia and location-based pricing
She cites examples of companies charging different prices depending on where a user logs in from, such as charging more for the same hotel room if the user appears to be in a wealthier area.
StubHub and fee variation
A test with production assistants showed that while base ticket prices were the same, fees varied by as much as $60. Owens notes that incognito mode reduced price dispersion, suggesting data collection affects pricing outcomes.
DoorDash and “agitation” scoring
She mentions a DoorDash patent for a multi-factor “agitation index” that could incorporate:
- battery life
- scrolling speed
- scrolling behavior
The implication: companies can use urgency and stress signals to adjust prices or promotions.
Amazon, Walmart, and agent access
The conversation also covers how major platforms may respond to AI shopping agents:
- Amazon reportedly blocks some agents like Muse
- Walmart is reportedly building infrastructure to register and authorize agents
This points to a future where large platforms control which AI agents can shop on their sites.
Broader Economic and Social Concerns
1. Pricing can reinforce inequality
Owens argues that personalized pricing may not just “soak the rich.” It can also target people who are under pressure or time-sensitive:
- a parent urgently buying medicine
- a traveler needing a ride home
- someone shopping while emotionally stressed or distracted
In other words, willingness to pay is not the same as ability to pay.
2. Price discrimination may overlap with protected characteristics
Even if companies aren’t explicitly discriminating by race, age, or gender, they can use proxies such as:
- ZIP code
- IP address
- geolocation
- device type
Owens cites a Princeton Review example where pricing appeared to vary based on location associated with Asian American neighborhoods.
3. AI may weaken price competition
She warns that many companies are using technology not to increase competition, but to suppress it:
- McDonald’s may use pricing algorithms to discipline franchisees
- landlords used RealPage-style tools to align rent increases
- gas retailers may be using pricing software to coordinate higher prices
Her concern is that these systems reduce real competition rather than improve it.
The Best-Interest Problem for AI Agents
Fiduciary duty for agents
A key policy idea in the episode is that AI agents acting on behalf of consumers should have a fiduciary-style duty:
- they should work in the user’s best interest
- they should not prioritize sponsored products
- they should not behave like disguised ad platforms
Owens compares this to financial advisers, where the law has long wrestled with whether advisers must act as fiduciaries.
Why this matters
If AI “agents” are really optimized for the platform or the highest bidder, then they are not truly serving the consumer. The episode argues that transparency about who the agent serves is essential.
Historical Context: How We Got Here
The rise of fixed prices
Owens gives a short history of pricing:
- haggling was once common
- Quakers and later retailers helped popularize fixed public prices
- John Wanamaker is credited with early price-tag retailing in the late 19th century
She argues the price tag was a kind of public good.
The airline industry as the pioneer of dynamic pricing
The airline industry is described as the original adopter of modern revenue management:
- limited seats created incentives for dynamic pricing
- deregulation pushed companies to develop pricing models
- by the 2010s, dynamic pricing spread more broadly to Uber, Ticketmaster, and beyond
Notable Quotes and Ideas
- “Pricing has really changed as we know it.”
- “This is a world of surveillance pricing.”
- “Who does your agent work for?”
- “If it’s calling itself your personal agent, it should probably work for you.”
- “The future of shopping, the future of pricing, [is] very much wide open right now.”
Bottom Line
The episode argues that AI is not just changing how people shop — it is changing the structure of pricing itself. Consumers may gain powerful tools for comparison shopping and automation, but businesses are deploying equally powerful tools to maximize extraction and exploit moments of urgency. The result could be more convenience, but also less fairness, less transparency, and a deeper shift away from one-price-for-everyone retail toward individualized, data-driven pricing.
Practical Implications
For consumers
- Be skeptical of AI shopping agents that may prioritize sponsored results
- Compare prices carefully, especially for travel, food delivery, and tickets
- Use incognito mode or limit data sharing where possible
For policymakers
- Consider whether AI agents need fiduciary-style obligations
- Re-examine pricing practices that may function as hidden discrimination
- Watch for algorithmic coordination that reduces competition
For businesses
- Expect more scrutiny over how prices are set
- Prepare for regulatory attention around transparency and discrimination
- Recognize that trust may become a competitive advantage in AI commerce