The 3Es of AI
As I write this (late December 2023), it’s been only a scant 12 months since OpenAI released ChatGPT. It’s hard to believe that, given how it’s caught the attention of companies, governments, and the general public and arguably become the fastest consumer software application to date. Heck, even my 80-year-old father calls me up to talk about ChatGPT!
Terminator or Saviour? You decide! (Image generated using Midjourney)
Outside a large and burgeoning start-up community building out AI products, several companies - Atlassian included - have introduced AI capabilities into their existing offerings. This raises the question of “where”, “when”, and “how” AI should be best used. There’s a lot of understandable concern from all walks of life about the threat of job losses, ethics, and what bad actors can do with the technology. I acknowledge these concerns but am opting to focus this blog post on how and where AI can boost the knowledge worker's productivity.
The way I see it, life is a series of tasks and decisions (title of my self-help book?!). In a work context, these decisions span from the mundane (“How do I complete this query about next year’s holidays?”) all the way to the strategic (“What sorts of products or services should we offer?”). There are broadly three domains in which these decisions are made:
| The Clear Domain - when faced with a decision, we ask ourselves “Have we done this before? If so, is there a routine response we can follow?”. This is the domain of bureaucrats (and don’t get me wrong, bureaucracy isn’t evil; it’s too much or too little of it that leads to problems). In this context, AI can help with pattern recognition and heuristic selection. Perhaps you’ve setup JSM’s Virtual Assistant in Slack to understand the natural language inputs about IT issues or HR queries and serve up the appropriate pre-programmed responses. In other words, AI enforces existing processes and policies. |  |
| The Complicated Domain - we find ourselves needing to investigate the situation further before we can find the correct response. This is the domain of experts. In this context, AI can help with information acquisition and preliminary analysis for a human expert to review and process. For example, it can automatically open an incident ticket and bring in information about similar incidents that might be ongoing. In other words, AI enables work to get done faster, and with greater consistency and accuracy. |  |
|
The Complex Domain - Complexity is a whole area of academic research, but suffice it to say that in this domain, there is no one person or group that can direct actions; the context in which decisions need to be made is always partially unknown; decisions themselves feedback and change how people behave (as an aside, this article is a good introduction to complex adaptive systems). Teams need to navigate their way towards their objective through a series of small experiments and using probes to understand the changes those experiments have created. This is the domain of tinkerers! In this context, AI can help by suggesting new ideas or even writing user stories to be reviewed by product teams (who might subsequently change how they think through and build their products). In other words, AI “creatively” enhances existing work. |
|
To sum that up, depending on the context, an AI can enforce, enable, or enhance work patterns - that’s the 3Es model. In reality, it’s more than likely that an AI might be doing two or even all three of those things. A virtual assistant might be enforcing the use of knowledge articles and enabling a human agent to quickly step in with all relevant information. It might be enforcing internal style guides and templates when writing long-form content while enhancing content by bringing in summary information from external or web sources. Or perhaps it might be enabling a post-incident review by summarising ticket information while enhancing the review by looking up publicly available information published by other companies.
What do you think of the 3Es model? How would you use it to shape your use of AI tools, and do you think there are more “Es” to explore? Let me know in the comments below!