In Part I we explored the idea of AI agentic systems, current paths being explored, and the roadblocks present in those paths. If the API-driven approach is “too hard” and the browser-driven approach is “too soft” does there exist an approach that is “just right?”
I’ve been designing and building data-driven & machine learning-enabled applications, on and off, for roughly 15 years. Although I was fortunate to have access to GPT3 about a year ahead of the mainstream, I was wholly unprepared for the wide-reaching impact of this technology. The introduction of ChatGPT in late 2022 brought into sharp focus just how impressive the current generation of large language models have become. The sheer breadth of zero-shot capabilities demonstrated by a single model has everyone’s minds racing. There is obviously immense power here, but how do we harness it to realize its true value and potential? That is the trillion-dollar question.
One of the most compelling visions for humanity’s next steps with AI is creating autonomous agents; AI systems that can do more than summarize text and provide plausible, information-shaped responses to our prompts. These are AI systems that can interact with other systems or the physical world. Although a great deal of R&D is underway as I write this, I’m shocked that one of the most promising solutions to the conundrum of how to achieve such agentic systems continues to fly under the radar…
REST is the architecture of the web and the web has seen unprecedented growth and evolution since its inception over 30 years ago. A web of static documents gave rise to the read/write web–the so-called web 2.0. New media types and formats have evolved, protocols have become more powerful and more secure. In short, the web has only grown bigger in scale, more powerful in terms of capabilities, and continues to evolve. The human web is a marvel of software engineering and its longevity is a testament to the vision its founders and architects. Unlike much of the software I use today, I have never opened a web browser to a message that the back-end of the web has undergone a breaking change and I would need to download new software before continuing.
At some point in the first decade of the 21st century the web crossed an inflection point; machine-to-machine API calls eclipsed human traffic on the web. Many of these APIs have been labeled REST APIs despite many–perhaps even most–of these APIs fail to exhibit the qualities elicited by truly following the REST architectural style. Consequently, the machine web consists of a lot of brittle integrations that require countless person-hours to craft and maintain. To be clear, I’m not here to get on my soapbox about how all these sinners are “doing REST wrong”; I really don’t care. Not every API needs to–or should–be a REST API. The REST architectural style is a very specific tool to solve specific problems.
“Some architectural styles are often portrayed as “silver bullet” solutions for all forms of software. However, a good designer should select a style that matches the needs of a particular problem being solved.”
-Dr Roy Fielding
Architectural Styles and the Design of Network Based Architectures
Fundamentally I want to talk about some of the overlooked aspects of this architectural style and how they can be implemented to eliminate problems with versioning, evolution, and flexibility. Today we’re focusing on content negotiation.
A few weeks ago I wrote an article on how investing in structured, semantic data can help move tools like ChatGPT from the “Trough of disillusionment” to the “plateau of productivity” and create intelligent agents that are actually intelligent. The core was that standardizing on REST Level 1 (or better) and beginning to layer in JSON-LD could provide a more meaningful and factual foundation for generative AI like GPT3 to deliver revolutionary value to organizations.
Less that three weeks later, Tony Seale, a Knowledge graph engineer, posted a brief demo video of these ideas in action.
Since ChatGPT became publicly available last November there has been an explosion of interest, articles, blogs, videos, arguments for-and-against; it can be difficult to separate out the hype from the reality. If you haven’t yet played with the public beta, it’s worth taking a look; first impressions are often downright startling.
One of the most impressive capabilities might be ChatGPT’s ability to seemingly answer questions asked in a casual, conversational manner and many hailed this as “the future of search” with Microsoft and Google both scrambling to integrate these capabilities into their search engines. A mind-bogglingly complex language model trained on a web-sized corpus of text boasts stunning capabilities although it doesn’t take long to discover that beneath ChatGPT’s impressive grasp of language there is a serious lack of knowledge. Google’s parent company, Alphabet, recently lost 8% of its market cap–roughly $100b USD–after their live-stream conference demonstrated Bard, their language model, returning incorrect answers.
The Mastodon post above summarizes my–and so many other’s–experience. Never in my 20+ years in the industry have I seen a technology move from the “Peak of Inflated Expectations” to the “Trough of Disillusionment” so quickly (see Gartner Hype Cycle). There is something powerful here, especially if it can be integrated with actual knowledge. Forward-thinking organizations are adopting the existing standards and architecture that just might be the key to unlocking the dream that GPT-hype represents. The first step may be as simple as evolving your API strategy.
This question, it would seem, has been answered countless times on countless blogs, articles, conference talks, papers, etc. yet here I am, joining the throng to tilt at this windmill.
My understanding of REST has been evolving continuously over the past 15+ years. I continue to find new nuances, new applications, new patterns, and rediscover concepts that I once completely misunderstood. I have brilliant friends and mentors, but I’m an autodidact at heart and more often than not, they merely illuminate the path. As I’ve sought to navigate this life and the information space we call the web, I have learned being self-taught is fraught with peril.