ChatGPT’s Disruption of Enterprise Architecture

The old state of EA, as assessed in 2022 by Forrester and in 2023 by Sparx, has been disrupted by “the elephant in the room,” ChatGPT. Sparx’s more recent poll, done this year, took place before the influx in activity around ChatGPT and last week’s AI-focused announcements by Microsoft at its Microsoft Build conference. We can now see the OpenAI and Microsoft strategy in full detail, built around the concept of a platform business model. It seems that Bill Gates and Satya Nadella learned long ago that a Platform Business Model and Open-Source strategy can pay off, as long as you provide the digital platform ecosystem’s API layer open to partners and developers and stimulate lots of new ideas around it. For Bill Gates it is now obvious that a platform can generate economical value for everybody who uses it when it exceeds the value of the company that creates it.

In a recent interview, Elon Musk who provided some initial founding to OpenAI, has expressed some astonishment that the OpenAI Open-Source approach has been transformed into a closed-source model, which means that Microsoft is investing into OpenAI, but on the condition of owning all key ChatGPT intellectual property. I am certain that this move involved a crowd of patent lawyers before the investment was made.

In light of  all these announcements, a key question I have is  how it will affect the state of our EA profession. When I look at Forrester’s recommendations from 2022 it seems that EA leaders had to reposition their EA approach to help their organization meet the ongoing need to digitally transform and respond to the COVID-19 pandemic. EA leaders  had to refocus on a technology strategy rather than on one influencing business strategy or business project delivery. Then when I look at Sparx’s poll I can see that, in summary promptedto me by chatGPT4 (after initial session of feeding and prompting “the beast”), there are many elements that I agree with or have already suggested to chatGPT:

  • The profession of enterprise architecture is evolving in response to various trends and challenges. The COVID-19 pandemic has accelerated the shift towards remote work, with a majority of architects now working in virtual teams across different geographies. This remote work environment requires architects to develop strong remote collaboration skills and leverage digital tools for effective communication and teamwork.
  • Architects continue to face difficulties in communicating the value of their role to executives and key stakeholders. The lack of physical proximity hinders their ability to demonstrate their contributions, emphasizing the need for enhanced communication strategies and techniques. Leveraging storytelling skills and design thinking approaches becomes essential for architects to bridge this communication gap and effectively convey the value they bring to the organization.
  • Traditional tools and methods are being replaced by digital alternatives, but there remains a need for more advanced architecture tools and capabilities. Integration of architecture tools with other IT systems, such as project management, finance, and solution delivery, is critical to accessing comprehensive and accurate operational data. Architects should also consider leveraging enterprise BI and risk management systems to gain a holistic view of operations and make data-driven decisions.
  • The emergence of AI technologies, including ChatGPT4, presents both opportunities and challenges for enterprise architecture. AI can augment architects’ decision-making processes, automate certain tasks, and provide advanced analytics capabilities. Architects need to stay updated with AI trends, understand the potential impact on their role, and explore ways to leverage AI tools effectively within their architectural practices.
  • As the profession of enterprise architecture moves forward, there is a need for talent development and succession planning. With a significant portion of the architect workforce approaching retirement, organizations must invest in training the next generation of architects. This includes nurturing skills in data analytics, AI, and other emerging technologies to ensure a skilled and adaptable workforce.
  • Overall, the state of enterprise architecture in 2023 reflects a continued evolution in response to remote work dynamics, the need for effective communication and collaboration, the adoption of digital tools, the integration of IT systems, the potential of AI technologies, and the importance of talent development. Architects who embrace these changes and leverage the available tools and technologies will be better positioned to drive successful architecture initiatives and support organizational goals.

As you can see, this version of knowledge in ChatGPT4 has not been fed yet by the recent announcement from Microsoft and doesn’t also represent any highly intelligent advice nor exhibit any “sentient” aspect of character (no starting to worry about poor EAs that need to adapt …).

The way I see things now is different than those stated in these two reports from 2022 and 2023. There is an elephant in the room and if we don’t want to be trampled by this new actor on the EA scene, we need to adapt and be very quick. I’ve begun studying AI already in 1989 (from the first book on this topic by Patrick Winston). At that time,  we didn’t have such powerful infrastructure and computing facilities, nor some of the algorithms invented later that allowed us to build Generative LLM (large language model) AI based on platforms capable of  storing huge amounts of information acquired from the Internet and trained by many contractors to reduce its hallucinations (to reduce the chance of providing invented information). Then in the 1990s, we began building rule-based expert systems capable of providing advice on specific topics such as taxation (i.e., this led to the development of an Expert Taxation System at the Australian Taxation Office).

As we are getting into the age of singularity, Digitalization and AI development can represent both threats and opportunities. In rereading Philip Kotler’s book, Marketing 5.0 ,published before ChatGPT went viral, I note some well-known elements on both sides of the balance (The Perils and Promises of Digitalization):

But Generative LLM-based AI will cause a tremendous disruption to many sectors and many types of jobs. We can already see in many areas where the “digital divide” is leading to the loss of jobs, while it is also true that it creates new ones. However, these new jobs require one to learn a new set of skills. Because the  AI genie has gotten out of the bottle, nothing will stop it now. Also, the current state of geopolitics will drive further AI development, including in the defense sector (i.e., AI-driven laser gun technology). In 2021 Daniel Kahneman, a Nobel Prize winner, stated that “Clearly AI is going to win. How people are going to adjust is a fascinating problem”. The European Community will attempt to muzzle it through new legislation that will be published this June, where risk management-based control will be initiated to help manage its use and proliferation. Hopefully, it will have the same impact as GDPR (data privacy act).

In light of the information summarized above, the following are notes on  how I see the impact of AI on our EA profession and the EA tooling that we’ve been using so far:

  • The segment of the aging EA workforce will have challenges to adapt since many in this segment use EA knowledge to provide income, while delivering limited business value to their companies. This has led already to the elimination of several top EA teams because their top management felt that they focused too much on high-level ,abstract activities, technical standards, and principles without getting their “hands dirty” with implementation teams, and thus not truly providing discernible value-added. They need to be directly involved with teams in business and developmental value flows going forward, which means that they must be able to wear several hats and to update their technical knowledge regularly.
  • I do not agree with Forrester or Scaled Agile (SAFe) that EAs need to focus mostly on technical challenges if they really want to help with Business and IT alignment. Instead, they need to understand what is driving the need for change to conduct more valuable architecture reviews of change or project proposals. They need to develop their storytelling skills and design thinking techniques to cross this digital divide and to be considered as a function providing demonstrable value.
  • EA tools must change. First, EA tooling needs to be better integrated with collaboration, project management, and service management tools for more collaborative and participatory architecting and flow management. We can foresee that ChatGPT-like platforms, with their additional plugins, will offer new semantic interfaces to EA tooling. With these,  we’ll be able not only to access structured information modeled in an EA Repository around its metamodel, but also unstructured information that resides in the current silos just mentioned.
  • This can lead to AI tools having access to a continuously updated knowledge base that will represent a “Living Digital Enterprise” that could be monitored through an  EA Dashboard, while also prompting architects and executives to act on topics highlighted by signals from the marketplace or leading indicators as defined in business and technology strategy and applied through a balanced score card approach. In the past, large organizations also used a foresight technique to develop strategic scenarios; in the near future, we can expect that periodic reviews of these scenarios will be fed from an AI platform where we would store and constantly augment our State of Knowledge about the organization. Such new approaches to information management will also be more accessible to smaller enterprises to better prepare them for upcoming disruptions.
  • LLM-AI based platforms allow for gigantic amounts of information to be stored (part can be public, while another part can be private information within a given enterprise). Such information could  be shared in real time by a number of specialized agents, each having specialized knowledge (developed by partners through plugins), while a human brain has its limitations in terms of short memory and an  inability to do parallel processing. Task switching does not lead to optimal thought process while ChatGPT has no such limitations and are now more efficient in storing and analyzing information than a human brain.
  • Finally, in consideration of AI’s possible wide-ranging applicability to EA, the new skill of “prompt engineering” must be mastered by EAs to still play a crucial role in this new environment.

Note: Your feedback is welcome. If you have some, please write to the author at the following address: [email protected].

By Alex Wyka, EA Principals Senior Consultant and Principal