There’s a lot of talk these days about how AI will reshape the mattress industry, upending the old rules and ushering in a dramatic new world, a smarter world.
Well, I submit to you that AI hype is running wild, outpacing the realities of what AI can do to help our industry. And I’ve got support for my view from AI itself.
In a revealing, exclusive interview, ChatGPT fessed up to a whole host of limitations.
“Overemphasizing AI in the mattress industry isn’t just a tech risk — it’s a strategy risk,” ChatGPT told me. “The danger isn’t using AI; it’s letting it drive the business instead of supporting it.”
Well said, ChatGPT.
My research assistant — that’s a helpful way of looking at mattress consumer-facing, generative AI — quickly listed nine ways that AI can lead us off the rails if we overemphasize it in the mattress industry.
“Commoditization of the customer experience” is No. 1 on its list. “AI pushes toward optimization — price, conversion, targeting. But mattresses are still a high-touch, trust-heavy purchase.”
Losing human trust leads to losing mattress sales, it adds.
No. 2 on its list of limitations is that AI sells what’s measurable, not what matters. AI optimizes for clicks, conversions and short-term demand signals, but mattress success often comes from comfort education, trial experience and long-term satisfaction, ChatGPT told me.
“AI can unintentionally push the industry toward: promotional pricing wars, feature inflation (cooling, tech gimmicks) and short-term wins over brand building,” it noted.
Still think AI is the be-all and end-all in Mattressville?
Bad data leads to worse decisions, ChatGPT said, adding that is a “huge” mattress problem. “Retail AI is only as good as its data — and retail data is messy,” it said.
By now, I think you get the idea. Lots of bad things will happen if we outsource our thinking to AI. That’s because, crucially, AI doesn’t actually think. Don’t be fooled by its fast-draw, detailed responses to almost any question. AI is great at sorting through mountains of data, pulling out pithy points and presenting them in a compelling, humanlike manner. As I said earlier, it’s a wonderful research assistant.
In a recent Wall Street Journal article, Christopher Mims offered a thoughtful take on “Why Even Smart People Believe AI Is Really Thinking.”
“Americans are becoming increasingly convinced that artificial intelligence is actually thinking like humans do,” he wrote. “This flies in the face of all we know about its inner workings: AI can’t think, doesn’t have a mind, and, in fact, is inherently untrustworthy.”
That article included a survey in which almost half of AI users said they think the AI models they are using are a little or a lot smarter than they are. And Mims quotes an AI expert as saying that AI models create “the illusion of consciousness,” distracting from the technology’s usefulness as “highly accelerated information processors.”
You might be thinking that chatbots are a great AI tool, but caution is warranted there, too. Chatbots are “fundamentally sycophantic,” one savvy mattress industry observer told me. “This makes them even less reliable for advice on most everything, including which mattress to buy.”
Still want to turn the keys of the kingdom over to AI?
Let’s adjust those unrealistic expectations about what AI can do for us. ChatGPT told me that the winners in the AI derby won’t be the most AI-driven, they will be the most AI-disciplined. We should use AI for augmentation, not substitution. And we should fix the business model before we automate it.
Don’t misread this column. I’m not saying we should ignore AI. But I am saying we should be smart about how we use it. We have real brains. AI doesn’t. At least not yet.
And, in case you are curious, ChatGPT says I know what I’m talking about.
“In the context of the industry, mattress writer David Perry (often referred to as Dave Perry) is widely regarded as a knowledgeable and respected voice in the mattress/bedding world,” ChatGPT told me.
So I’ve got that going for me. Thanks for the kind words, ChatGPT. Now you get back to your high-powered data processing and let us humans think about the best ways to put you to work for us.
