Remember blockchain’s brief, bizarre moment in the cultural spotlight?
Sotheby’s sold an NFT collection for just shy of$17 million. Big-name companies like Coca-Cola were appointing in-housechief metaverse officers. A virtual Gucci bag sold on Roblox for over$4,100, more than the cost of its physical counterpart. Matt Damon and Tom Brady were shilling FTX, the crypto company whose CEO would soon be imprisoned for conspiracy and fraud. You could barely spend more than five minutes on the internet without seeing a Bored Ape something-or-other.
Hubris is always clearest in hindsight, obviously, and there are many aspects of the metaverse/crypto/NFT craze that today look… well, crazy. That’s especially true of the lingo and arcane mechanisms that grew around the decentralization subculture, a phenomenon immortalized in a May 2022 tweet from NFT marketplace RareCandy: “a lotta yall still dont get it,” the tweet read. “Ape holders can use multiple slurp juices on a single ape… so if you have 1 astro ape and 3 slurp juices you can create 3 new apes…”
Whether it was written tongue-in-cheek or in earnest, the post became a symbol for the absurdity of the NFT/crypto craze, with its arcane terminology blurring the boundaries between memeculture, video games, and finance. Years later, it was definitely right about one thing: A lot of us still don’t get it.
https://x.com/rarecandyio/status/1521548615579226114
The same can be said about the modern
The same can be said about the modern AI industry, which, like the NFT sector of yesteryear, is getting tangled in linguistic and technical cobwebs of its own making.
In their all-out race to monetize the technology, AI companies have branched chatbots into a million different tools, each with its own capabilities and intended audience. For all their talk of building “super apps,” developers have for the most part focused on building standalone products—the idea being that winning over customers is largely a matter of casting as wide a net as possible. New models, meanwhile, have been released nearly on a weekly basis and given multihyphenate names that probably appeal to the data-driven minds of software engineers but for the rest of us are rather unwieldy.Since September of last year, OpenAI has launched ChatGPT Pulse, GPT-5.2-Codex, GPT-5.5, GPT-Live, OpenAI Presence, GPT-5.6 Sol, Terra, and Luna, GPT-6 Astra, GPT-6 Sol and Luna, and ChatGPT Space, among other releases. In a single day last month during its annual developer conference, the company unveiled over twenty new products, including Decisions API—a product thatappears to be closely modeledon TypeSafe’s Jev—anddots, always-on AI agents that come with Pixar-esque avatars, developed in direct response to Meta’s Muse.
Okay, deep breath. If you felt like you were losing your mind reading that last paragraph, I assure you that’s a perfectly sane response to an AI industry relentlessly bombarding all of us with AI products. It’s hard to read through that menu and not hear echoes in your head of the infamous “slurp juice” tweet.
Making matters even more confusing, many of these
Making matters even more confusing, many of these products come with their own Byzantine pricing structures and usage limits that seem to shift with the wind. (No wonder many people and businesses have been canceling their ChatGPT or Claude subscriptions in favor of open-source models from China, privacy risks be damned.) Customers have been responding. Anthropic has had to fend off a class action lawsuit alleging it misled its customers about the token allowances afforded by its Claude 5x and 20x subscription plans. And following an announcement from OpenAI thatit was halving the usage allowance provided through its $200/month Pro plan, some power users took to social media in protest.
https://x.com/kimmonismus/status/2105648440323534947
“[Customer] feedback is clear that you all want things to get simpler,” Thibault Sottiaux, head of ChatGPT and Codex at OpenAI, wrote in a SundayX post. “On it.”
The AI industry is very much still in an experimental phase as it works to turn AI into a product that individuals and businesses are willing to consistently pay to use, so, naturally, there’d be something of a Cambrian Explosion in how those tools are commodified and sold. This is almost always the case with new consumer products: witness the folding, slide-to-open, and BlackBerry-style keyboard phones that proliferated before the iPhone ushered in the now-ubiquitous touchscreen.
For AI developers who are already struggling
But for AI developers who are already struggling to justify their monumental spending sprees on a technology whose long-term profitability is still very much an open question, the confusion of product names and prices has become a liability. A report from consulting firm Bain & Company published last week found that by 2031, the AI industry will need an annual revenue of $6 trillion in order to stay afloat and cover its enormous computing costs. If it has any hope of hitting that lofty goal, it needs all of us—not just uber-tech-savvy power users—to enjoy the AI tools they’re building enough to pay for them. That’s a long, long way from happening.
As Sottiaux suggested in his X post, companies would be well advised to focus a little less on shipping as many products as possible, and a little more on consolidating their menu of existing products, and establishing clear, jargon-free channels of communication around technical updates and pricing changes.If AI companies don’t want to go the way of NFTs and the metaverse, they need to start speaking a language everyone can understand.
