Navigating the AI revolution: A journey through 3 phases of understanding
It’s been a confusing few months for everyone involved with digital technology. Whether provider or consumer, we are all struggling to make sense of the latest wave of AI-driven digital advances. Will their automation and intelligence displace large parts of existing ways of working? Do we rush to adopt them or wait to ensure all the safeguards are in place? Are they friend or foe? Yet, the one thing that we can all agree on is that it’s going to take some time to get to grips with AI and its implications.
In thinking about this recently, I was strongly reminded of how I felt more than 30 years ago as I began to engage people about the foundations of the Internet. That too was a difficult time of transition as we moved from a growing base of standalone computing capabilities to an interconnected set of digital products and services supporting shared facilities to expand their use.
How did we make sense of that dramatic shift? Looking back on this time, is it possible to distinguish the distinct steps in this journey?
Many years ago, when the Internet was something yet to be so embedded in all of our lives, I remember the trouble I had trying to explain it to people. In this first stage of the Internet, the use of computers was largely limited to so-called “back office” tasks taking place behind the scenes in areas such as accounting, stock control, and complex engineering tasks. Public discussion of the future of computing was primarily a technology debate for engineers and computer scientists. For everyone else it was irrelevant or unfathomable.
By the early 1980’s, many people now involved with the growing wave of software development and the application of computers were becoming excited by the possibilities we saw for wider access to more powerful technology. The beginnings of the Internet. It would bring a universal infrastructure connecting every computer in the world. We’d agree on the protocols and standards that support a global platform for integrating applications of every kind. Soon everyone would be able to access the digital plumbing required to interact and share information across an infinite web of data. Such statements were well meant, but caused as much confusion as they created clarity. Few people could get their heads around what was happening to get any sense of the disruptive effects this would have.
Then, in the second phase, when email addresses and webpages started to become more widely popular, explaining the Internet became so much easier. The era of the WorldWideWeb had arrived. Here was a use of the Internet that gave insight into the possibilities provided by digital technology that everyone could grasp. Rather than a focus on the power and possibilities provided by the internal engine of the Internet, no matter how extraordinary that was, a simple usage model emerged that made it much more real for many people.
Now the challenge had shifted from explaining the intricacies of the technology toward helping people to get online. Few understood the mechanisms that lay beneath the surface, the world of Internet Protocol addresses, Universal Resource Locators, domain name servers, hypertext transfer protocols, and so on. For me and others, it resulted in many hours explaining where to register your own domain name, how to configure a domain name server, how to write your first lines of code in hypertext markup language, what to do if your email stopped arriving, where to create and host webpages for greatest impact, and so much more. The key was understanding enough of the context to unlock the mysteries of aligning the different components to ensure all the pieces were configured correctly. And that wasn’t easy.
Eventually, of course, all of this work was increasingly simplified and automated in the third phase. A wide range of tools and services were created to automate much of this work, check that the details were correct, and provide user-friendly interfaces to allow on-going maintenance and updates. So that eventually free tools such as WordPress made web hosting much easier, and companies such as Wix, Squarespace, and GoDaddy made webpage creation cheap (or free) and simple. As a result, access to the Internet grew significantly as many more business and individuals used the WorldWideWeb to advertise, interact, and conduct their activities.
The new focus was less on how to use the Internet, and more on how to apply it appropriately. The most important questions became focused around the content being created, the audiences being addressed, and the ways that could be used to influence information exchange and commerce using the Internet-based capabilities of the WorldWideWeb. We had now progressed to worrying more about the right ways to deploy these solutions, their effects on products and services, and how this new paradigm was redefining current ways of working.
It is no surprise then, that this same 3 phase journey can also be applied to the journey we are now pursuing with AI. While some people have been deeply engaged with its development over many years. For most, the arrival of AI as a meaningful personal and business issue is very recent. And it is the impact of simple use cases brought by generative AI that has made the difference to why this is relevant today.
The first phase of AI was dominated by an internal focus on AI technology. It was the domain of engineers and computer scientists looking to understand and improve its performance. Incredible energy was devoted to this work over many years. But for all but a few, this was hidden from view. News stories about winning at go or chess games aside, little was seen of these advances.
More recently, the second phase brought several use cases forward, most significantly the widely applicable “question-and-answer” solutions based on generative AI systems that are now widely available. While they represent just a small part of the world of AI, the emergence of ChatGPT and other easily-accessible tools provided a wake up call that AI-based solutions can readily be applied today. As millions rushed to play with tools like ChatGPT, the promise of this form of AI hit home.
The key has been the straightforward nature of the use cases that people quickly understand. In many cases, their use is seen in chatbots querying large knowledge bases, text generation to aid a wide variety of writing tasks, real-time language translation, and simple ways to manipulate text and images. These have quickly captured the imagination of many individuals and companies. A lot of playing and piloting has taken place. In some cases, capabilities have begun to be embedded into existing products. However, much of the initial excitement has turned to more serious questions about their appropriate use and the challenges of applying them effectively.
As a result, the third phase has begun. Increased automation and simplified tools have emerged to reduce the overhead and increase the impact of generative AI-based solutions. Many of them are little more than new interfaces on top of the generative AI engines offered by OpenAI, Microsoft, IBM, and others. Even so, with a few clicks they help people to use these rather complex capabilities to ingest a large dataset, train a large language model, configure an attractive interface for users, and embed these capabilities into the workflow of typical tasks.
Many companies are now releasing products and services to support the application of AI to carry out different tasks. These are now widely available in domains as diverse as sales management, health, education, customer service, and legal services. Whether it is small startup solutions such as chatnode for creating data-driven chatbots or large companies such as IBM releasing its enterprise-ready watsonx solutions, we will be seeing a very large number of products and services aimed at simplifying the adoption of generative AI solutions in the coming months. This will bring even greater access to this technology to a wider set of individuals and companies.
In a confusing digital world, it is important to step back to reflect on the path we are travelling and assess current directions. In doing so, we can see that the journey of understanding and adopting AI technology mirrors the phases of development of the Internet. Initially, confined to tech experts, AI’s accessibility with tools like generative AI systems such as ChatGPT that engaged a broader audience in real-world applications. Now with simple use cases such as chatbots, text generation, and language translation, a new phase is unfolding, marked by increased automation and more user-friendly interfaces. Many companies are releasing exciting AI products for various sectors like sales, healthcare, education, and customer service to simplify AI adoption, making it accessible to a wider audience. Just as the Internet evolved from complexity to widespread use, we are now seeing the wider effects of AI on our ways of working.
