thingsithinkithink

AI Demystified Book by Antonio Weiss

Pearson FT published AI Demystified offers a gentle introduction for business leaders who want to understand how AI might impact their field.

David Gérouville-Farrell avatar
  • David Gérouville-Farrell
  • 6 min read
Key frameworks for practical AI implementation

Because I’m obsessed with AI, I go to as many AI-related events in London as I can. I blagged a ticket to the launch of Antonio Weiss’s “AI Demystified,” which was held in the swanky Financial Times Pearson building on the Strand. After getting a copy of the book, I discovered that a project I’d worked on was featured in it. Naturally, that gave me an extra incentive to actually read the thing.

AI Demystified book cover Antonio Weiss’s AI Demystified book launch

But beyond self-aggrandisement, I think there are a few valuable bits worth sharing from the book. This is especially true for people who are less deeply enmeshed in the AI world and are wondering how they can position themselves and their organisation to adapt successfully to this rapidly changing landscape.

Horizontal vs Vertical AI Transformation

Weiss provides a useful framework for thinking about the scope of AI implementation. He distinguishes between horizontal and vertical approaches:

Horizontal transformation involves changing entire end-to-end workflows. His example is Octopus Energy, which uses AI to handle over a third of customer emails with an 80% satisfaction rate. The AI manages complete customer interactions from initial contact through resolution, with productivity gains equivalent to 250 staff members. This approach also provides scalability during demand spikes, such as meter reading days when enquiries increase significantly.

Vertical transformation focuses on enhancing expertise within a specific domain or professional area, rather than changing the overall process. According to research from Imperial College London, AI tools can detect up to 13% more cancers than human radiologists in mammogram screenings. The AI doesn’t change the overall diagnostic workflow but significantly enhances one crucial component.

Weiss’ Verb Exercise to Prioritise AI Opportunities

If you haven’t brought AI into your organisation yet, one way you might want to do so would be through what Weiss calls the “verb exercise.” The approach is straightforward: for any task you’re doing - writing, researching, interviewing, generating concepts - list the active verbs that consume the most time. These might include:

  • Imagining new landscapes
  • Writing interview guides
  • Summarising interviews
  • Searching for information
  • Generating concepts

The next step involves asking how generative AI might assist with these time-consuming activities. It’s straightforward, but it can bring your attention to good candidates for AI automation without getting lost in low-value opportunities.

Page from AI Demystified showing case studies including FCDO project The FCDO Consular Assistance Project that I worked on

We used a slightly different methodology, but basically that’s why the FCDO Consular Assistance Project was the first use case the Foreign Office focused on. The amount of time spent and the number of members of staff whose entire days were filled with answering routine email enquiries meant that it was a good candidate for bringing AI in. That project is now launched across all UK Embassy pages internationally, and they’ve seen a drop in 80% of the incoming written enquiries. Getting the use case right is the single most important part of bringing AI into the organisation.

Weiss’ AI Principles

One section that’s quite useful is Weiss synthesising a set of AI principles based on those from major international frameworks. The collection alone is quite good to see:

From these frameworks, he distilled his own set aimed at business implementation:

  1. You have a specific problem you are trying to solve - Start with user needs rather than AI capabilities
  2. You know what good looks like - Define success metrics that improve on current performance, not perfection
  3. You understand AI capabilities and limitations - Recognise what generative AI can and cannot do in your context
  4. You act responsibly, legally and ethically - Navigate the evolving regulatory landscape appropriately
  5. You maintain control - Ensure humans can explain, switch off, and trace AI decisions

One of my engagements just now is with local authorities who are asking me to help them put principles and governance in place for the use of AI, and I think these above are a pretty decent starting point.

The Emerging Chief AI Officer Role

Weiss explores the emergence of Chief AI Officers (CAIOs), citing David Salvagnini’s appointment at NASA as one of the first such roles. Early adopters include GE HealthCare, UnitedHealth Group, Deloitte, IBM Automation, and Mayo Clinic. I noticed last week that the UK Government is seeking to appoint their first ever Chief AI Officer as well.

He breaks down the CAIO role into five key capabilities:

  • Strategic vision setting: Scanning opportunities, understanding current capabilities, and setting future AI direction
  • AI assurance: Auditing, reviewing, and maintaining continuous oversight of AI tools and systems
  • AI implementation: Executing delivery of major AI projects and programmes
  • Skills development: Understanding organisational capabilities and developing gap-bridging plans
  • Compliance: Maintaining oversight of evolving regulatory and legal landscapes

How to Stay Current

One thing that I get asked a lot is how can non-AI obsessed people possibly stay up to date with this dramatically changing landscape. In the book, Weiss calls out a whole bunch of resources that are quite useful and I think I may start referring to these in future. Unfortunately though, there really is no shortcut apart from spending time, energy and attention.

People to Follow:

Newsletters and Mailing Lists:

Professional Networks:

Research Sources:

Tools and Platforms:

thingsithinkithink

  • The CAIO role breakdown feels about right for where the market is heading. As AI becomes more central to operations, having dedicated senior leadership makes sense, though the specific responsibilities will likely evolve rapidly.

  • Notably, he doesn’t include the UK Government’s AI Playbook principles, which offers 10 detailed principles specifically for public sector AI implementation. There are a few things in the UK principles that are quite valuable and not present in this section of the book, particularly Principle 5 around understanding the full AI lifecycle - that’s important if you’re going to be bringing these things into your organisation.

Recommended for You

Why we need Experiment-based Roadmaps in the AI Product Era

Why we need Experiment-based Roadmaps in the AI Product Era

Why evaluation-driven experimentation creates better roadmaps in AI products.

Is Psychology or Politics Behind Project Failures?

Is Psychology or Politics Behind Project Failures?

How Big Things Get Done (Ch2): Exploring the forces that drive us to think fast and act slow