We thank Denis Rothman for taking part in the Data Science Interview Series. Denis Rothman is an Artificial Intelligence Expert and is the author of several popular books such as Transformers in Natural Language Processing, and Hands-On Explainable AI. In this interview, Denis Rothman shares his academic background and research interests. He also shares valuable insights stemming from his decades of experience in the field of AI, including:
- Changes in the AI Landscape over the years
- Advances in AI Research
- Journey to Becoming a Best-Selling Author
- Importance of Writing in the Learning Process
Academic Background and Specialization
Bala: Could you tell us about your academic background and specialization?
Denis Rothman: In the 1970s and 1980s, I studied Linguistics, Civilizations, and Computer Science at Sorbonne University and Paris-Diderot University. When I graduated, I taught Computer Science at Pathéon Sorbonne. I started my freelance Linguistics and AI activity while I was a student. During that period, I registered a “word2vector” patent that gained university (post-graduate degree), media, and corporate recognition. I obtained public funding to continue my research. In 1986, I registered what is now a “chatbot” patent that led me directly to LVMH (Moët et Chandon division) and Airbus (formerly Aerospatiale division).
My professors were cross-disciplinary and encouraged exploration. I traveled to many countries to understand how civilizations were built, how languages were structured.
I became familiar with Supply Chain Management (SCM) which led directly to AI SCM projects. I rapidly specialized in Artificial Intelligence through a passion for algorithms.
Changes in the AI Landscape over the years
Bala: As an expert in the field of AI, could you share your insights on how the AI landscape has changed over the years?
Check out what books helped 20+ successful data scientists grow in their career.
Denis Rothman: In the 1980s, we worked on personal computers or small mainframes (servers). We did not have the Internet, so we had to physically go around the world to meet people and pick up new ideas. This explains why I traveled so much during my college and early freelance years.
In the 1980s and 1990s, you had to have a solid reputation for the word to go around to deliver AI optimizing algorithms for Supply Chain Management successfully. I focused on optimizing, so I made it fine during those years.
Starting in 2006, cloud servers became available, and the Internet reached a good level of maturity. It changed everything! No more traveling so much, no more installations. We could sell SaaS (Software as a Service). The users would have to connect with their browsers, test for free then pay a monthly license! On top of that, we could access good papers and documentation from around the world. We could also share our ideas.
When artificial intelligence became a mainstream trend around 2015, we were ready, had the solution and the experience. So, a fantastic journey began and is still continuing!
Advances in AI Research
Bala: What, according to you, are the most significant advances in AI research in the recent years?
Denis Rothman: The following three trends are mind-blowing:
- The arrival of powerful Artificial Intelligence models trained with supercomputers such as OpenAI’s GPT-3 engines. These models are industrialized using structured transformer models, for example. The texts they generate cannot be distinguished from human expression.
- The embedding of Artificial Intelligence everywhere, from our smartphones to our social media activity. Google Search has improved tremendously provided snippets of the search answer without clicking on a link, for example. Google Translate is progressing daily. Everything in AI is evolving very fast!
- Industry 4.0 Artificial Intelligence, which is used to connect everything to everything, everywhere. So from China to the US, from Europe to Africa, the world is being connected physically and digitally from the Americas to everywhere!
Bala: What do you think are the challenges in adopting AI at scale in the industry?
Denis Rothman: There are two key factors beyond the classical project management constraints encountered during any project:
It is challenging to explain that Artificial Intelligence is only a tool. It’s not organic. It is not a living organism. AI is only math. You can either be the tool of AI or use it as a tool. It is also tough to explain that Industry 4.0 is generated millions of micro-decisions in the supply chain flows around the world. Without AI, there would not be enough humans to face these challenges without slowing the world down slower food delivery, vaccine production, clothing, and everything we need daily.
- Cross-disciplinary knowledge
To be able to fully understand the requirements of a project, cross-disciplinary knowledge is mandatory. For example, Linguistics for NLP or notions of law for applications of AI in governance and other legal applications.
Bala: According to you, what are the areas in AI that offer some of the most exciting research opportunities?
Denis Rothman: I love Artificial Intelligence because I love algorithms. The most interesting aspect of any AI project is creating algorithms. Programming is another exercise. Programming is translating an algorithm into code. Programming comes with constraints, bugs, machine power, and criticism from others.
But when you are designing an algorithm in the middle of a quiet night, it’s like communicating with the structure of the universe. It is like writing a music score, drawing, or painting. You are creating something out of nothing but yourself!
Computer vision, Natural Language Processing(NLP), and all of the AI-ML-DL algorithms are fascinating to design! It’s like playing different music scores.
Journey to Becoming a Best-Selling Author
Bala: As an author of several books on popular topics in AI, such as Transformers, Explainable AI, could you please tell us about your journey as a technical author?
Denis Rothman: I love Artificial Intelligence, people, and sharing. I began writing about ideas, logic, and reasoning when I was in high school. Sharing has always been exhilarating for me. When I share an idea, the person will react and ask difficult questions. In turn, this forces me to think harder and explain better. After my ideas are structured, we both see the light together!
Sharing is enjoying togetherness in the society. You do something, you analyze the reaction, and you adapt.
If you do not share, you do not learn. If you do not share, you cannot understand what you are doing because you need a third party to tell you what they see in your ideas beyond your confined ecosphere.
I wrote a lot of documentation for corporations and some studies. However, in 2017, Tushar Gupta, a visionary working at Packt Publishing, encouraged me to share my ideas, knowledge, and experience through books. Packt has a unique ability to create publishing teams. You are never alone. So you can share your work and get a lot of feedback before anything is published. That is fantastic!
Importance of Writing in the Learning Process
Bala: What is the importance of writing in the process of learning AI? I strongly believe that writing reinforces understanding and the learning process. What’s your opinion on writing tech tutorials when learning a particular concept?
Denis Rothman: There are basically 4 approaches to language:
- Understanding through sound
The words cross your mind at full speed. It’s difficult not to miss a lot of content. You can get information through prosody (intonations and tone of voice, for example). However, in the world of technology, content matters more than intonation.
- Speaking through sound
When you speak, you are compelled to be clear, or nobody will understand you. However, you don’t have to worry about punctuation, the length of sentences. Also, unless you are reading a text, you are going at full speed without controlling the content.
Reading is no doubt deeper than listening. You can capture quite a lot of information and think about it for as long as you wish. You are picking up a lot of ideas.
Writing is the ultimate form of communication. If you did your research correctly, it’s the sum of the three previous dimensions! First, you have listened to others. Second, you have made an effort to explain yourself through speech. Third, you have picked up a lot of information by reading others. Now, you can compile everything you feel and know into the content you can share! You have prepared quite a work of art for your readers. You are now eager to know what they think and start a communication process at another level.
Bala: What are some of the challenges associated with using AI in Healthcare?
Denis Rothman: Privacy in Healthcare worries me. We are not machines, robots, or things. We are human beings! We deserve respect and the ability to keep our dignity. We are not lab rats.
That being said, medical progress relies on statistics! Using AI to find patterns is undoubtedly necessary.
I think that government-led ethical committees should decide what is good or bad for us, not private sector investors. That is about all I can say as a citizen because I’m not a Healthcare expert! ?
If you were to suggest a few books that can help AI enthusiasts to gain strong foundational skills, what would they be?
Denis Rothman: The tempting answer is “mine”. ?
I’m going to disappoint you because I do not believe in flashy knowledge. Artificial Intelligence is there to help machines make decisions for us about our culture. To design effective AI, we need to be educated.
First, I would recommend learning to review Mathematics in detail: Arithmetic, General Algebra, Calculus, Geometry, Trigonometry. I would then go in this order among many others: Plato, Aristotle, Descartes, Pascal, Kant, Nietsche, Levi-Strauss, Boltzmann, Poincaré, Markov, Lyapunov, Einstein, Shannon, and Turing. Then, the army of AI authors, yours truly included!
Bala: Could you suggest a guest that you’d like to see in this interview series?
Denis Rothman: There are many great names I could recommend! However, I deeply admire Margaretta Colangelo. She is one of the greatest pragmatic thinkers of this generation and deserves more recognition.
Bala Priya is a math, programming, and data enthusiast with a keen interest in technical writing and editing. Her areas of interest and expertise include Python, Machine Learning, and Natural Language Processing.