Giannandrea Downplays The Significance Of AI Chatbots

Bloomberg:

Giannandrea’s purported lack of urgency may be as much philosophical as temperamental, though. Conservative about the pace of AI development and skeptical about the value of chatbots, he’s argued internally that there’s little urgent threat from OpenAI, Meta, Google and the rest. Instead, colleagues say, Giannandrea maintains that what users want in an assistant is an interface for controlling devices. Despite the delays and setbacks, he still holds to that vision.

Maybe the market has already superseded the philosophy, but I can understand the aversion to a chatbot experience. As long as you have a display, users don’t want to do the work of the system. The system should lead them onwards. That means user interfaces; buttons, sliders, icons. Chatbots present an open-ended textbox and leave everything else up to you. Until we get to the era of mind-reading, user interface elements are going to win out over textboxes. It doesn’t necessarily mean human curation. Maybe AI models will end up building the perfect custom UI for each situation.

However, the technology behind chatbots does not feel antecedent. It feels like the future. And a text field lets real people access that futuristic technology (the underlying power of the LLM) right now. The name chatbot is also a bit reductive. I use ChatGPT to get tasks done. You ask it to do it, it does it, you move on. This is where these things have utility. The term chatbot implies ideas of para-social conversations and pleasantries with robots. ChatGPT will certainly confabulate to infinity and simulate human-like interactions, if you approach it that way, but it isn’t really where most users are finding value in the product.

I would agree with the notion that “what users want in an assistant is an interface for controlling devices”, but I think it is pretty clear that LLMs already get pretty down far that path, if you give the frontend user a text input field and give the backend model access to enough system APIs.

So for screen-based devices like computers and phones, a chatbot textbox certainly seems compelling for the present. It makes Apple seem way behind on AI — even more behind than they are — when in lieu of a chatbot, they seemingly employ that argument to justify shipping nothing at all. Apple exacerbated this issue further by shipping UI that looked an awful lot like a chatbot app, with the new Type to Siri UI under the Apple Intelligence umbrella, despite not actually shipping anything like that.

The anti-chatbot philosophical take breaks down further when you consider other types of devices. For devices that don’t have screens, or have small enough screens that voice becomes the preferred input method, I don’t know of anything better than a chatbot. The Apple Watch, Apple TV, and HomePod lines depend on voice input, and if you wired up a chatbot to the other end of the dictation pipeline, it undoubtedly results in a better experience. The open-ended textbox is quite naturally your voice. Modern chatbots give reasonably good answers to almost any questions you can utter to them.