Companies continue to try to make AI a product, not a feature

Key Takeaways

• The hype around AI has led to a flood of futile attempts to make it the “next big thing” in various fields.
• Initially, AI will not be a standalone product but will act as an enhancement, supplement, or productivity adjustment.
• Companies need to focus on successfully integrating AI into existing products and delivering meaningful value without forcing a paradigm shift.

While AI has been talked about in one form or another for years, the wave of investment that has swept into the industry in the wake of ChatGPT's November 2022 release is unprecedented in the technology industry. It has been years since we have seen such a wave of hype around a new technology in a similar fashion.

But as the hype begins to die down and actual products get into consumers' hands, one thing is becoming clear: AI is far from the “do-it-all” killer AGI that some have promised. The inevitable inflation of expectations and the proliferation of free money have led to a flood of truly awful attempts to make AI the “next big thing” in everything from washing machines to pin-on assistants.

In the short to medium term, AI won't really be a product. It will be an enhancement, a supplement, a nice-to-have suggestion, or a productivity tweak in our lives. No one is underestimating ChatGPT. But AI needs to show its value in small ways long before general tools become more useful.

The sooner the hype cycle ends, the better

AI fandom is out of control and everyone wants a piece of the pie

Despite the end of the so-called “free money” era brought about by ultra-low interest rates, AI hype is still reaching new heights. The tech industry is full of companies seeking to pioneer AI, touting everything from home assistants to pharmaceutical diagnostic tools, all powered by AI (in theory).

Everyone wants to find the next unicorn, the next Snapchat, Uber or Meta, but sifting through a never-ending mountain of pitches and product ideas is like trying to find £210 million worth of Bitcoin in a landfill – you've got a lot of rubbish to sift through.

For consumers looking at this space, it's mainly been the experience. Outside of kickoff offerings from the likes of OpenAI, most AI products or integrations have been primarily garbage. Even companies like Microsoft and Google have faltered along the way.

Dedicated AI products have failed

This is by no means a universal trend, and AI “integrations” like Copilot have fared better than dedicated attempts to build AI-centric products. Some of the more specialized options have fared better, with some success stories like perplexity.ai and ChatGPT, while hardware products like Rabbit R1 have failed spectacularly.

Monetizing AI is Hard

This AI rush has one defining characteristic of the early 2010s tech rush: no one has yet figured out how to monetize it. ChatGPT probably has the most solid model: a monthly subscription with priority access to the latest models. But building/training models comes at a cost for companies to broker access to the datasets. In the future, we expect to see products launch where the hardware heavily subsidizes the software, with rug pulls and subscription requirements. The harsh truth, recently demonstrated with Spotify's Car Thing, is that cloud services won't last forever and companies are willing to sacrifice post-sale customer satisfaction for profits.

AI can improve products, but it should be used sparingly

Apple might have the right idea

There is light at the end of this tunnel of high expectations. Apple's WWDC AI (pronounced Apple Intelligence) series of announcements was a welcome one. Many small enhancements, most of which are done locally, seem like a very sensible approach. Such soft integration allows users to maintain their existing usage patterns and avoids unpleasant integrations (such as Copilot) that are forced on users, while leaving room for later improvements.

It also reduces the blast radius of potential damage from hallucinations or misbehaving models. At worst, it can result in inappropriate suggestions for simple actions. Hopefully, something like the Google AI suggesting you put PVA glue on pizza won't happen here.

Companies like Google and Microsoft aren't going away anytime soon, and neither are the operating systems, apps and platforms they control across the consumer tech landscape, so perhaps they could follow Apple's lead and gradually integrate in ways that are very useful.

Companies should focus on making AI a feature, not a product

There's a lesson to be learned here: Companies that are successfully leveraging AI are focused on integrating AI in a meaningful way with existing products, rather than shaking up the entire ecosystem. Elegant AI integration, like the ML integration we've seen in everything from photos to search over the last decade. This allows them to deliver meaningful value to consumers without forcing a paradigm shift.

This also helps prevent the inevitable “burn-out” felt by customers, as the constant bombardment of everything with AI can be exhausting and lead to resentment towards the industry as a whole. A good example of this is smart home appliances – the push of the Internet of Things era to add intelligence to everything in the home: lights, ovens, refrigerators, washing machines, etc. Many consumers loved these, but they often generated more resentment than excitement.

AI doesn't yet have a clear “product”

Another thing to consider here is that, apart from the obvious (chatbots have certainly demonstrated their value), AI doesn’t yet have a clear product. Obviously, everyone wants to discover what the killer product for AI will be, but restraint in focusing on research rather than shipping junk would be justified.

I'm excited about the AI-enhanced future

Although I've been pretty light on AI here, I remain excited by its potential. AI will provide smart automation, intelligent sorting, and helpful suggestions. I look forward to fast access to information and indexes, fast translations, and the ability to quickly edit, manipulate, and identify objects in images and videos in real time. I also look forward to the day when the hype dies down and AI becomes a practical, useful tool, seamlessly integrated into our lives. Then maybe we'll start to see AI as a true “product.” Until then, I hope companies take their foot off the gas and start thinking about smaller, smarter, slightly more useful implementations.