The promise of Artificial Intelligence (AI) has captivated imaginations for decades. Today, with every new development, it seems we're on the cusp of a life-altering AI revolution. Yet, if you were to venture outside and ask 100 people if they engage with AI technologies like ChatGPT every week, you'd be hard-pressed to find more than 20% who say 'yes.' So, despite the hype, why hasn't AI become as ubiquitous as, say, the smartphone?
AI is incredibly proficient in specific tasks but lacks the wide-reaching, autonomous capabilities that would seamlessly integrate it into our daily lives. At present, AI is like a knife without serrated edges; it can cut, but not as efficiently as we would like.
Take ChatGPT plugins as an example. These plugins can connect with different APIs to carry out tasks such as retrieving live weather or searching for flights. But herein lies the crux: much of the web isn't just a series of APIs awaiting a command, most APIs are hidden away through layers of authentication and bot detectors waiting to show you 4 photos of buses or traffic lights to click.
Most websites today aren't static information; they are dynamic, interactive platforms that require nuanced interactions. When you book a flight, for instance, you don't just send a single command to an API. You choose dates, compare options, fill in personal information, and more. These steps require 'clicks' and 'taps,' something that current AI cannot perform autonomously.
An AI equipped to handle such intricate tasks could navigate the labyrinthine spaces of the internet to perform a multitude of functions without needing your direct input at every step. This would mean that the AI could handle complex chores like shopping for groceries, comparing insurance plans, or even completing governmental paperwork online and do some without having to reinvent the internet we have today.
The technology is not far off. Recent advancements in machine learning and computational power I think gives us enough language understanding through LLMs for this to work. It is now just a matter of time before a developer or team create a library for layering on top of a click-and-tap interface.
Once AI learns to 'click and tap,' it may finally live up to the monumental hype that has surrounded it for years. And 2024 might just be the year where all these elements come together, turning the promise of AI into an integral part of our daily lives.