In today’s episode, I will comment on five lessons learned comprised of an AI Playbook Methodology developed by DeepLearning.ai intended to avoid AI digital transformation initiative pitfalls.
Let me start by sharing with you a reflection: any seasoned project manager will tell you that the following is a best practice: (1) Documenting the current situation (baseline), (2) Documenting the expected outcome (goal), and (3) Build a project plan (WBS) to comply the stakeholder expectations in time, at cost, and with controlled risks.
While attending the AI For Everyone Course by “Deeplearning.ai” via Coursera, Andrew Ng (a well known AI leader) shared with the audience a different approach, based on his own experience, that I would like to share with you.
During the course, it introduced the following AI Transformation Playbook:
- Execute pilot projects to gain momentum
- Build an in-house AI team
- Provide broad AI training
- Develop an AI strategy
- Develop internal and external communications
I. Execute Pilot Projects to Gain Momentum
As you may see, the first step is not to build a grand vision or to create a super strategy but to build an informal team and expose them to AI-based pilot projects first.
His rationale is closer to Lewis Carroll (Alice in Wonderland) saying:
“If you don’t know where you are going, any road will take you there.”
Andrew Ng commented that customers often force a vision or strategy without a good understanding of the actual capabilities of AI.
In addition, he recommends choosing pilot projects following the Goldilocks Principle: (an analogy of the Three Bears children’s story). The proper scope should not be simple or complex but just the right size.
In this context, the Goldilocks Dilemma points us to the need to choose an initial project where we have a real possibility of success and credibility in 6 to 12 months. This is critical to allow the AI team members to acquire the experience and skills to provide effective feedback to the oncoming initiatives.
Another lesson learned:
The best approach is to look for opportunities for Task elimination or process optimization instead of focusing on job reductions to find financial returns.
Why? Because the risk of automating a task is much lower than the risk of (incorrectly) reducing a job position.
Side story: Recently, I had a sad experience with an airline reservation: I bought a ticket at a premium price because I knew my trip might be canceled, which did happen.
I was able to cancel the reservation and ask for a refund via the portal, but I never got a confirmation email, so I called their call center to learn about the claim status.
To my surprise, the call center was utterly automated (bots). Sadly, it didn’t include any option to ask for claim status: it took me two weeks to contact a supervisor (another story) who told me it would take 30 days to get the money refunded.
During the following weeks, other issues were found that “impeded” the airline from completing the refund. I had to call them several times, following a complicated non-standard procedure, to get in contact with a supervisor for follow-ups.
Sadly, I am still waiting for the money back after three months.
II. Build an In-house AI team
Disregarding if the Digital Transformation will be executed by internal resources or by Systems Integrators, it is essential to build an AI team comprised of AI SMEs and Business Leaders. They will work together, looking for good opportunities for AI-based initiatives.
III. Provide Broad AI Training
This methodology’s primary goal is to reach a certain level of competency of all critical stakeholders before building a vision, strategy, or plan.
Once we have a couple of success stories, the organization typically is ready for cross-pollination: other organizations / BUs will want to work with the AI team, and by doing that, the following approach is to level the understanding of Business Leaders and IT SMEs in what AI is, what AI is not, and what are feasible projects they can deliver together with few approvals to check their capabilities and commitments.
The rationale is that AI is a disruptive technology, and the right approach is to help stakeholders understand its value and limitations so they can make informed decisions.
So, the main recommendation is to build medium-complexity projects of incremental value and complexity. By doing these, the whole team will mature and deploy massive initiatives on time, at cost, and with controlled risk.
IV. Develop an AI Strategy
Did you realize that Developing an AI Strategy is the fourth step of a five-step playbook?
This is a significant difference from the traditional project management best practices.
The rationale is that as AI is a disruptive technology, this playbook documents several activities before building the vision so that the critical Stakeholder will make decisions aware of the capabilities and limitations of AI.
The way I see it, this playbook includes an initial phase zero that strives to make all stakeholders competent and informed of AI capabilities and limitations.
It also documents that this approach is not the only one; at times, customers utilize a different sequence of tasks with varying levels of success, but in general, this sequence of activities provides the highest successful implementation index.
V. Develop Internal and External Communications
Sure, any success story must be communicated, any risk must be shared in advance, and lessons learned should be collected.
A closing statement:
The course I attended is projected to be completed in four weeks. It covers many other areas of AI: Artificial Intelligence, Machine Learning, Data Science, Building AI Teams, Ethics, use cases, successful approaches, and limitations.
I recommend it to any potential AI Stakeholder, Business Leader, or IT SME.
Good enough?
What are your thoughts on the subjects raised in this edition of the Digital Acceleration Newsletter?
I hope you find them valuable and pertinent. Your feedback is appreciated.
Share them in the comments below, and if you have ideas about other topics you’d like to see covered in this newsletter, feel free to add those suggestions.