AI is going to destroy (some) OSS architects and consultants. Does Good Will Hunting provide an antidote?

It seems fairly obvious (to me at least) that AI is an existential threat to many traditional information-based roles like consultants and architects. I’ve already seen one wave of this happen, so the next seems inevitable too.

When I first entered the field of IT and telco consulting back in 2000, our clients engaged us because we had knowledge of, and/or access to, information that they didn’t. We also found ways to synthesise that data into reports / recommendations that our clients could action.

However, as more information has become available online in the years since, clients have become more adept at doing their own research. The “General Practitioner” consultant business model became less necessary to clients, so this type of consultant has already needed to evolve.

The next wave is already rolling in. Generative AI solutions not only have access to vast amounts of data but are also good at synthesising the data into reports. They’re even beginning to reason with the data. AI tools are only going to get better at this too, which puts many information-based roles at risk. Unfortunately the human expertise required to perform consulting and architecture roles is no longer as unique / exclusive as it was before Gen AI.

That’s daunting for an OSS architect and consultant like me. It is an existential threat!

As a result, I’ve been giving a lot of thought to how the industry might evolve in the years ahead and how PAOSS will adapt to add value to clients in evolving ways. I feel that the answer might just lie in a scene from the movie, “Good Will Hunting” starring Robin Williams and Matt Damon, but we’ll get to that shortly.

Clearly data gathering, synthesis and report generation has/will become semi or fully automated. If that’s all you/I do, then our relevance is likely to diminish (being honest, it already has 😉 ).

But when you think about it more deeply, those three highlighted deliverables are rarely ever what our clients want anyway. They rarely want data collection or a report. They do generally want what’s either side of those points though. Let’s highlight this in a “customer lifecycle sequence”

1. Understand and articulate a customer problem (empathy)
2. Access to primary data
3. Ask questions of data
4. Gather data
5. Synthesise / Assemble data
6. Create a report and recommendations
7. Develop an Implementation Plan
8. Implement to solve customer problem
9. Operate and Maintain (to continue to solve customer problem – if applicable)

AI currently excels at the highlighted middle three activities, but is already pushing upwards and downwards into the list above. Just as the AI landscape is evolving at an barely comprehensible rate, so is my interpretation of how to remain relevant into the future, but here goes:

The most important activity in this list is number 8 – the ability to implement. This introduces a credo that I believe strongly in and becomes even more important to consider for anyone who finds themselves doomscrolling:

“Be a Producer, not just a Consumer”

That is, build stuff. Learn how to implement. Experience the challenges of implementation. Learn how to overcome the obstacles of implementation. Do!

Once you have developed the experience of implementing, it unlocks the other items on the list that aren’t highlighted.

#7. Once you’ve implemented, it makes you more credible at planning future implementations

#8. Once you’ve implemented, you develop the knowledge and skills to implement again and/or guide others through the challenges of implementation

#9. Once you’ve implemented, you better understand ongoing operations and refinements that can be made to drive continual improvement

#1. Once you’ve implemented and operated, you have greater understanding of the problems that customers face. You also have a greater empathy for what the client is experiencing, the fears, the soft-skills of communication, persuasion, relationship building, etc that are always necessary to cut through pre/mid/post implementation challenges

#2. Once you’ve implemented and operated, you’re more likely to have gathered important data, insights and relationships that simply isn’t accessible by an AI

#3. Data alone doesn’t tell you anything. You have to know the right questions to ask of it. Invariably, having deep implementation and operation experience helps you to ask more pertinent questions. As Alex Hormozi has stated,

“Due to tools like Google and Generative AI, we now have the power to know almost anything we’re curious about…”

Whilst algorithms / machines / neural-nets can certainly help to automate many technical implementation activities – data aggregation, issue diagnosis, decision-making and even some implementation or operations tasks – they don’t (yet) have the deeply human elements of implementation experience.

This is where Matt and Robin come in on an iconic scene in Good Will Hunting:

So, here’s how I’m thinking:

• Don’t just think, talk and report; do, experience, implement (it’s one of the reasons I believe the OSS sandpit series is an important exercise for me personally)
• Build soft skills and relationships
• Continually learn and upskill
• Focus on solving bigger and more complex problems – step up the value chain (from OSS product/project-level problems, to cross-suite, to company-wide, to intra-company, to industry, and beyond)
• Be the human co-pilot for companies that haven’t already been blessed with the opportunity to implement OSS/BSS solutions 🙂

If you’re in an OSS-information-based role, what are you thinking? How are you handling the existential threat that faces us? How are you seeking to evolve? What are your mantras?

Please leave us a comment below.

Hat-tip to David Ziembicki, who wrote this article, which provided the Good Will Hunting link to AI and provided inspiration for this blog.