From Code Whisperers to Chatterbots: The Evolution of Human Computer Interfaces
Today, we stand at the threshold of a new era in which conversational AI and natural language interfaces (NLIs) promise to redefine how we interact with technology. Human computer interface with technology will fundamentally change in the LLM era. We might see several headless assistants for workflows or products that are more push based surfacing the insight at the right time instead of when a user initiates action. Clippy might’ve just been ahead of its time.
Act I: MS-DOS and the Command Line Conundrum
Picture this: It's the early days of personal computing, and MS-DOS reigns supreme. But there's a catch—no pretty buttons, no colorful screens, just a black void with a blinking cursor. Want to copy a file? You better know the right spell—err, command. It was like a secret club, where only code whisperers could speak the lingo. The masses yearned for something more user-friendly.
Act II: GUIs to the Rescue—Enter the Age of Clickable UIs
Lo and behold, the graphical user interface (GUI) made its grand entrance, and computing was forever changed. Windows, icons, and menus transformed screens into digital playgrounds. With a trusty mouse sidekick, users could point, click, and drag their way to computer wizardry. While Xerox PARC was the original developer of GUI, it was Apple's Macintosh that unlocked its true potential. Computing was no longer a dark art—it was a visual fiesta!
Act III: The Rise of the Chatterbots—AI Gets Conversational
Fast forward to the present day, and there's a new kid on the block—generative AI and language models like GPT-3. These AI maestros can chat, write, and even crack jokes (well, they try). User experiences have gone from typing commands to having full-blown conversations with digital beings. "Hey Siri, what's the weather like?" "GPT-3, write me a quirky article!"
And the best part? These AI buddies are getting smarter by the day, thanks to machine learning and heaps of data. They're here to help with everything from booking appointments to penning poetic verses.
This is nothing short of revolutionary. GUIs will potentially be replaced by a much simpler natural language interface (NLI) or the NLI will be augmented by some limited GUI elements. For example, the next generation of modern data stack will be truly self serve (albeit sometimes with a human) or products like hu.ma.ne that are building Star Trek communicator badge will become real (no wonder the founders of Humane were builders at Apple).
Product teams will start to care about the training data because it will have a significant impact on the user experience. Consumers will need to get used to products that are more probabilistic than deterministic, as well as how to communicate with these new systems. Expecting users to immediately adapt to this probabilistic world will be an error and will need to be well thought into the design of new applications.
This will require evolution in how we design products and experiences. A lot of the product experiences today involve user generated actions. We would move to a world where the right knowledge and action is surfaced at the right instances to the user, thus reducing the overall work by a user.
Conclusion
In conclusion, this is supporting the case for startups. When the AI-driven user experience is a major upgrade compared to the old generation of the product, it opens the door for startups to step in and displace the incumbent. But products will need to be thought grounds up from first principles. Founding teams will need to rethink the product and design aspects given the probabilistic nature of the technology we are dealing with.