Thomas Davenport and Nitin Mittal’s 2023 book, “All in on AI,” makes the case that large companies can uncover significant value by going “AI first” or, as they say, “AI Fueled.”
I found the book to be a fast read. It may be best for large legacy businesses and organizations.
The book does a good job of compiling the latest thinking about creating an AI organization. It is full of practical tips and case studies. I will keep this handy as a good reference.
Here are five of my takeaways:
You need to create a factory for building and deploying AI applications.
I found many good tips for putting AI solutions into production and adding value.
I think some of the exciting ideas in this area include using technology to deploy the solutions faster, to monitor the applications (there will be drift and the need to upgrade), to reuse algorithms (too many organizations spend too much time building the same thing multiple times), and to have applications to monitor and clean data.
For organizations just starting, I like the advice to pick multiple projects (some will fail), start with small wins (and build on these), and don’t be obsessed with short-term ROI (this can kill too many innovative ideas too soon).
Education is important.
Throughout the book, the authors urge you to consider educating your organization. “Smart companies are reskilling and upskilling their employees to develop, interpret, and improve AI systems.” And they go further to say that “Executives need their own version of AI upskilling.”
Be creative.
The book urges you to look beyond using AI for just operational efficiency. It also asks you to consider creating some new or using it to influence customer behavior.
It is obvious to ask how AI will help achieve your business strategy. However, to unlock creativity, they also suggest a company should have a strategy for becoming AI-Fueled. I’ve found that when more people in the organization understand and think about AI, more creative uses are found.
Borrow Ideas.
The book is filled with case studies—from short anecdotes to deeper dives.
For example, they mention that Shell used drones and deep learning to inspect pipes. Previously the task took six years (I’m guessing this was done manually). With drones and deep learning, it became a few days. Knowing the technology exists makes you look for these opportunities.
They have a long chapter in the book going through different AI applications by industry. This section can give you ideas for your industry and allow you to see opportunities outside your industry.
The deep dives are enjoyable. One of the authors works at Deloitte, and they go into how Deloitte uses AI. They also dive deep into Capital One. This is interesting because Capital One was already committed to analytics1 and still needed to reinvent itself with AI.
Not all the stories and cases will resonate with you, but there are enough that many will.
My complaint.
My one complaint with the book is the same with two of my favorite enterprise AI books, Competing in the Age of AI and Prediction Machines: All these books ignore (or minimize) optimization and Operations Research (OR).
This book mentions OR in one place, but it doesn’t make its long list of AI technologies.
For organizations committed to making better data-driven decisions, optimization and OR should be a big part of your toolkit2.
If you’d like to subscribe to the blog for free, click on the button
Capital One appeared in the 2006/7 HBR article and book “Competing on Analytics.”
I include optimization and OR as part of the AI. If you look at my four definitions, it fits into Practical AI. I will soon release another blog post on Practical AI.
Do you have any book recommendations that fulfills that element on the practical application of OR?