When I read the press release about Target partnering with ChatGPT for a shopping experience, I was intrigued. Here was a real-life example of AI shopping agents serving as your personal shopper from beginning to end. Or so I thought.
The reality was a hot mess MVP that delivered the worst of ChatGPT with none of the seamlessness I’d hoped for. The AI recommended products inappropriate for my needs, suggested items that weren’t in stock, and when I tried to purchase, it simply dumped me onto a standard Target page. So much for living in the future.
Why This Matters Beyond Retail
As someone who studies digital customer experience in financial services, I was interested in this for two reasons:
First, is this transactional use of an LLM the first step toward how we’ll interact with financial platforms? Will we soon be having natural language conversations with our bank or brokerage instead of navigating dropdown menus?
Second, and more critically: will consumers who’ve experienced clunky AI shopping tools trust an LLM with their actual money? Not $8 for frozen sweet potato fries that Target doesn’t stock, but their checking account, investment portfolio, or insurance claim for their crashed car?
What Good Actually Looks Like
Here’s what I expected: I’d mention to the AI that I’m going camping, and it would suggest realistic gear I might need. It would present a thoughtful list with enough context for me to make informed decisions. I’d approve the purchase, and the items would arrive at my door.
As we know from user research, what people expect is often far from what they actually want.
What I actually want is more nuanced: Show me what I might need, let me indicate what I already own, then walk me through the options for each item I need to buy—explaining the differences between products Target actually has in stock so I can make an educated choice.
The Financial Services Parallel
I suspect financial services product managers are watching experiments like Target’s closely, trying to understand how conversational AI will reshape their industry. The questions they’re grappling with mirror what Target is learning:
- How do customers want to interact conversationally with complex financial products?
- When should the AI make recommendations versus simply providing information?
- How do you balance automation with the need for human judgment on consequential decisions?
- What happens when the AI makes a mistake with someone’s money instead of their shopping cart?
The Trust Problem
Here’s what furrows my brow with all of this: trust is harder to rebuild than to establish. There’s currently some openness to trying these new AI tools—people want to believe in the promise. But when ChatGPT recommends out-of-stock products or inappropriate items, just another broken experience, it erodes that willingness to experiment.
How long until users are willing to risk their valuable time trying it again? More importantly, will a frustrating Target experience make consumers mistrustful of AI at their bank?
Financial institutions have spent decades building customer trust around security and reliability. They can’t afford to squander that goodwill on half-baked AI implementations that suggest the wrong investment strategy or mishandle a claim.
Financial institutions have spent decades building customer trust around security and reliability. They can’t afford to squander that goodwill on half-baked AI implementations that suggest the wrong investment strategy or mishandle a claim.
What Financial Services Can Learn
The financial services industry would be wise to watch Target’s integration carefully—not just for what works, but for how customers react to what doesn’t. Every poor AI experience in any domain raises the bar for what financial institutions will need to deliver before customers trust AI with their financial lives.
The potential for conversational AI in banking is enticing. But the path from shopping cart failures to retirement account success requires learning from these early stumbles. Financial institutions that rush to deploy AI without addressing the fundamental challenges Target is facing—accuracy, inventory (or product) knowledge, seamless transactions, and genuine understanding of user intent—risk poisoning the well for everyone.
My plea for financial services: Don’t launch something until you have something that works. Financial services can’t move fast and break things; you have to be the grownups in the room.
Adam Hulnick is the Vice President of user experience research at CI.