Podcasts supported by eBooks
Mastering Artificial Intelligence is a daily podcast show supported by eBooks which contain the key ideas covered in the podcasts.
The objective is to provide a complete, joined-up picture of the entire AI space while delivering the deep insights needed to really understand the field.
Mastering Artificial Intelligence is structured as 11 Modules each of which focuses a specific aspect of AI:
- The big questions – new thinking on AI that takes us closer to the truth
- Neuroscience – most of what you know is probably wrong
- Theoretical foundation – key insights that many programmers lack
- Technical building blocks – 10 key ideas that define how AI works
- AI in action – 200 ways AI is being used to solve real problems
- Creating the AI market – the main actors driving the market
- The leading edge – how state-of-the-art AI systems actually work
- Building stuff – how to use AI in completely new projects
- The economic impact of AI – markets, the economy and people
- AI in the future – the plausible, the implausible and the plain eccentric
- Fact vs. Fiction – taking on the Silicon Valley glitterati
Each of the above 11 Modules will consist of a set of around 20 podcasts of between 30 and 60 min each. Each Module comes with a specially written PDF eBook (premium content) of between 100 and 200 pages which you can take away and consult later.
The eBooks are not verbatim transcripts of the podcasts, but are specially written in an informal, conversational style while being structured in the form of a management report, with an Executive Summary, Introduction and Conclusion etc. Although the podcasts and eBook for a given Module contain the same ideas and basic content, they are significantly different.
Each podcast also comes with its own PDF eBooklet (premium content) of between 5 and 10 pages which contains the key ideas conveyed in the podcast. The eBooklets are written in the style of a white paper or extended blog article and typically contain images, graphics and other content that is not in the podcast.
Volume of material
In total Mastering Artificial Intelligence will include around:
- 250 podcasts, each between 30 and 60 mins
- 250 eBooklets, each between 5 and 10 pages
- 11 eBooks, each between 100 and 200 pages
Later on, when the base content of Mastering Artificial Intelligence is complete – which I think will take until the end of 2019 – I envisage creating a new Module that will include a range of long-form discussions with people who are expert in their fields.
I envisage that the style of these sessions will be to deeply discuss a particular controversial idea that arose while researching the programme – such as whether human intelligence might be quantised down to the level of individual neurons, whether biological software exists in the same way as computer software exists or whether the Tree of Life idea – which is the center piece of evolution by natural selection – might actually contradict Gödel’s first Incompleteness Theorem.
The focus of these discussions will be to build upon some of the ideas that have come out of creating Mastering Artificial Intelligence and will take the listener off the beaten track into new intellectual territory.
The eBooklets and eBooks contain links to books, research papers and online learning resources that were used in the research phase.
In addition, I sometime reach out to people by email to check thinking on a given topic or to float my own ideas to see what the reaction is. Mostly, the people I’m interacting with are senior-level academics who work in AI, mathematics, neuroscience and other related fields.
I should say that the approach taken with Mastering Artificial Intelligence is to try to create something that adds to what already exists – in a complex field where there are a lot of conflicting viewpoints.
If you resolved the AI field into a set of key ideas – drawn from right across the space – then that set of ideas would contain many contradictions. In other words, it would not be self-consistent.
To be blunt, some of the ideas in the AI space make sense, some do not and others are simply wrong. The challenge is to figure out which is which.
If the goal is to distill from this alphabet soup of ideas a subset that is self consistent, and then to further develop that into a rich narrative – which is the purpose of Mastering Artificial Intelligence – then you are going to have to make hard choices.
You cannot just write down what an expert says and accept it as fact. And you cannot sugar coat things or hedge your position with words ‘possibly’, ‘might’ and ‘could’. There is a place for content that is safely hedged like this does, but not here.
Instead, you need to calibrate what a given person is saying using your own deepening understanding of the field and, if necessary, reject their idea as being inconsistent with the facts.
But at the same time you also have to be prepared to accept that a carefully-constructed argument, which you formerly felt sure about, might be wrong.
You have to be willing to adapt and admit that you were wrong. But you also need to have a sufficient level of self confidence to push back and call something out if it doesn’t make sense.
So I do rely on external sources partly as a source of new facts and ideas and partly as a way to calibrate and enrich a gathering narrative.
Daily podcasts and eBooklets will begin to be published in July 2018.
Considering that I originally had the idea for Mastering Artificial Intelligence in Jan-18, you might be wondering why there’s still no content?
Part of the reason is that I need to think about production consistency: the sheer workload involved in creating a daily podcast and the associated eBooklet 5 days a week is such that the content you will be publishing on a given day must have been already researched.
I’m not going to be just talking randomly about AI, but offering a very carefully-considered position that is part of an overall, rich narrative which will frequently contain new ideas, new ways to look at familiar questions and many intricately constructed logical arguments.
Creating content like this takes literally 1000s of hours of work, or at least it does for me.
Given this, if you think about the reality of maintaining a consistent publication schedule you actually need about 3 months of content pretty much all complete and ready to go – before you actually start publishing anything.
As of June 2018, I’ve got Module 1 complete, Module 2 is in a very advanced state and Module 3 is coming along.
The plan is to continue researching the content for future Modules while almost ‘turning the handle’ while I publish content that has already been researched.
The other part of the reason is that as I’ve tried to explain before, AI is a very hard topic to wrap your head around: you really do need to look at a given question from several very different perspectives (e.g. metaphysical, software, math and neuroscience) before you can arrive at a position that you can support with a strong argument based on fact and logic.
This takes time, and that’s coming from someone who spent years running a technology industry research company, has a science and technology background and is used to the mental labor needed to understand complex topics.
In hindsight, I’m very pleased that I decided to wait until my position on certain contentious topics stabilised. For instance, the question of whether AI is real and whether AI as it is presently defined can ever be real are not questions you can answer by reading a page on Wikipedia. I’m pleased that I waited because I’ve changed my mind on both of these topics – after having looked carefully at the facts from all sides.