The uses of artificial intelligence 2018
In October 2017, I published my second ebook on the uses of artificial intelligence. Downloaded over 32,000 copies, it was obviously appreciated by all those who wanted to get a practical reading of what AI is and what can be done in companies. In short, the practical-practice outside the beaten track of anxious and peremptory refrains on AI or endless discussions on the advent of the singularity in 2045.
The discipline of writing had allowed me to create the contents of the two-day extension training that I give about four times a year to CapGemini Institute as well as to create the substance of my many lectures and interventions on AI.
All this encouraged me to put the cover back and create, a year later, an update of the book here. Here is the 2018 edition of the ebook “The uses of artificial intelligence”! This edition is a fairly complete overhaul of the 2017 version as you will be able to judge. Already, in terms of pagination, the absolute scarecrow for readers drenched in tweets and other short messages. We thus go from 362 pages to 520 pages. And I had to decrease the size of the font used so that it did not exceed 600 pages!
So here is the cover and the link to download this ebook in PDF format. No epub this time around due to the document’s condensed layout.
And since the PDF is 47 MB, here is a other link allowing to recover a more compressed version of 18 MB which will fit in memory in the readers of certain e-readers.
What justifies this overweight? It is linked to the method I use to create my works and that I will reveal to you. I call it the “squirrel-hazelnut” method. It consists in recovering and putting aside the topical contents on the subject to be treated then to unstack the whole at the time of the writing of the book. I complete this with a fairly intense plowing of open data on each new subject: startup databases, scientific databases such as Arxiv, retrieval of courses from major universities around the world, analyst reports as well as various video content. I thus collected more than 1500 documents to create this edition of the ebook. It’s far from perfect and there are obviously holes in the hazelnut fillet.
We can classify the novelties of this edition in four main categories:
- The AI science and technology : with a broader and updated overview of AI algorithms, particularly in deep learning, with differential programming, capsule networks, transfer networks, XAI, differencial privacy, homomorphic encryption, the evolution of the market for neuromorphic chipsets and GPUs, quantum machine learning and a little more background on data management.
- The horizontal and vertical markets : with the addition of 6 vertical markets, construction, utilities, telecommunications, luxury goods and fashion, services and consulting, education and a horizontal market, accounting. This part of the eBook has doubled in size, reflecting the diversity of AI solutions and case studies across industries. None escapes it.
- On the social issues with a focus on studies on employment, on the components of the ethics of AI and on the geopolitics of AI.
- On theadoption of AI in business with in particular a new section on intellectual property of AI.
Here is all of this broken down by topic. The part on the actors has been reduced because I have distributed a good number of pages describing startups in the previous parts. In blue, the pagination of the 2017 edition and in red, the pages added in the 2018 edition.
I deal laterally with many debates around AI: pAI ours and fantasies (there are too many, and not enough positive fantasies because both lack imagination and when it comes to fears, human civilization faces much greater threats before a possible general AI see the day), is it AI or not AI? (AI is a polymorphic technology, a toolbox, let’s stop thinking about it …), AI bias (the bias of data, algorithms, people and also the big bias of the rearview mirror consisting of relying too much on data from the past to predict the future and perpetuate the past), the place of French startups (good not good? Let’s stop comparing ourselves with the USA or with China. We are an intermediate country. AI does not change much to the position of the French and European digital industry) and skills and adoption (this is the main concern for companies and you don’t learn AI in half an hour).
This large document is still structured in eight main parts that are here and even better segmented than in the previous version:
- History and semantics of AI: what is AI? Who created the discipline? Where does this name come from? Why does no one agree on the meaning to be given to it? How is AI segmented from a technical perspective? What are its main intellectual currents? How has this new discipline progressed since the 1950s? Why has it had two great winters and what explains the current dynamics? Is it sustainable? Where are we today? How does AI compare to human intelligence?
- AI algorithms and software : what are the main mathematical and algorithmic building blocks of AI? Automatic reasoning and expert systems and why are they less talked about than in the 1980s? What are the techniques and applications of machine learning, neural networks and deep learning? Are recent advances coming from software, hardware, or data? What are the tools for developing and building AI applications and why are the majority open source? How are artificial intelligence bricks progressing? What about generalized artificial intelligence? Is it a fantasy? Can we easily reproduce the functioning of the human brain? What are the projects going in this direction and can they succeed?
- AI data : what is the role of data in AI? Where do they come from? What open data can AI use? What is data bias in AI and how is it avoided? What are the sensors that feed the AI data?
- AI hardware : what are the material resources that drive AI forward? How is the application of Moore’s Law evolving? Why are GPUs and neuromorphic processors now being used for AI applications? How do they stand out and classify them? Who are the players in this market? Why is there a big difference between training an AI and executing it in the consumption of material resources? Will quantum computing have an impact on AI? What is the role of sensors and connected objects? How are AI cloud resources managed as well as on the embedded systems side? How to architect AI solutions taking into account developments in processors, telecommunications, energy and security issues?
- Generic AI applications : what are the generic and horizontal applications of AI, in image processing, language, robotics, marketing, human resources, accounting as well as in cybersecurity?
- AI business applications : what are the major applications and case studies of AI according to vertical markets such as transport, healthcare, finance, insurance, industry, distribution, media, tourism, agriculture, legal professions, public services, defense and intelligence? Added to this in this edition: utilities, education, construction and real estate, luxury goods, services and consulting and generic Internet. Why are some of these markets more dynamic than others? How startups enable companies to innovate in these different markets?
- AI players : what is the strategy and what are the AI offerings of extended GAFAMI, including IBM, Google, Microsoft, Facebook, SalesForce, Oracle and many more? How do some of these actors deploy vertically? How are startups developing in general and then those of the French ecosystem in particular? How to assess the added value in AI of startups and other players in the ecosystem? How are AI solutions marketed? How much is product and how much is service and data?
- AI and society : views and studies on the potential impact of AI on employment, professions and on society in general. What are the limits of the predictions? How to avoid being robotized? How to prepare for the skills level? What are the main lines of the impact of AI on AI politics and policies in France and elsewhere in the world? What is the state of the geopolitics of AI? Is China going to invade us with its AI?
- AI and business : how can companies integrate AI into their strategy? What are the good methods and practices? How to manage skills? How to benchmark AI solutions? How to organize? How to integrate AI into other dynamics of digital-related innovations? How will the developer profession evolve? How to train in general?
As usual, I must thank the many people, researchers, entrepreneurs, IT managers of companies that I was able to meet and who provided me with various insights into the reality of AI today, all like those who participated in the proofreading of the book: Benoit Bergeret, Françoise Soulié Fogelman, Antoine Couret and Dimitri Carbonnelle.
Since November 2017, I have been delivering popularization training in the uses of AI as part of a two-day course organized by Capgemini Institute, here are the dates to come on Paris : November 15 and 16, 2018 (sold out), December 10 and 11, 2018 (sold out), May 16 and 17, 2019, October 1 and 2, 2019, November 18 and 19, 2019, and for the first time at Lyon April 9 and 10, 2019 (synopsis and registration). The program is a summary of the content of the ebook and updated regularly. I will also intervene on December 4 in the Cap Digital Campusfor a one and a half hour presentation on the uses of AI (registrations).
I also decline the content of this ebook in all imaginable formats for popularization in conferences, seminars and brainstorming workshops of variable format, training from half a day to three days, with a possible specialization in a large number of vertical markets (health, transport, banking, insurance, legal, construction, utilities, etc.). Writing is good for those who have time! But the oral completes well for those who do not have enough! In duet with Fanny Bouton, I finally decline the theme of AI by putting science fiction and science in abyss. We did it in Bordeaux on the occasion of The Great Junction in March 2018 as well as HEC in September 2018.
Like the others, this ebook is only distributed electronically. You can print it as needed on your own or by using the myriad of online printing services that are available. I would probably update it to correct any faults I discover or that you point out to me.
Good reading !
Olivier Ezratty is a consultant in new technologies and author of Opinions Libres, a blog on digital media (digital TV, digital cinema, digital photography), and on entrepreneurship (innovation, marketing, public policies, etc.). Olivier is an expert for FrenchWeb.