Getting to Know You: Internal Corporate Communications Interview with Jean Bolot
In this edition of Getting to Know You, Corporate Communications interviews Jean Bolot, InterDigital's Head of AI Labs, to explore the department’s latest news, unveil InterDigital's very first mobile app, and reveal surprising answers about the best AI Companion.
Corporate Comms: Jean, as Head of AI Labs, you lead one of InterDigital’s most cutting-edge and experimental departments. Following last year’s historic events, do you think that 2021 will be a year for AI? Did the events of last year change our relationship or need for AI algorithms and innovation?
Jean Bolot: For the past few years it feels like every year has been “the year of AI,” and rightly so because AI certainly has been a megatrend of the decade. But you are correct, and I think AI will still be in the spotlight in 2021.
Frankly, I don’t know how much AI has been used in the fight against COVID – was it a plus, or really a critical piece? But independent of that, there were several noteworthy AI advances in 2020. These include the development of ethical AI to address issues of bias in data or algorithms; the use of deep learning to discover Halicin, a brand new type of antibiotics, or to understand how proteins fold – which is one of the most fundamental problems in biology with dramatic implications for how our cells and tissues operate, and the release of GPT-3, which is a new type of giant neural network called transformers. These transformers are 2000 times larger than the networks today and are revolutionizing the processing and automated generation of text.
CorpComm: AI Labs had an eventful year! Looking ahead, what priorities does your Lab have in the exploration and application of AI and deep learning algorithms? What challenges are you looking to tackle this year?
Jean Bolot: Not surprisingly given the technology positions of InterDigital, our main focus is on developing AI-based advances for wireless and for video.
Much of our work on wireless is done with Jim Nolan's lab. For example, we are looking with Ghyslain Pelletier on innovations for 3GPP such as channel state estimation and compression using autoencoders, a type of neural network, and with Miki Beluri on beamforming. We are also investigating the use of transformers. I mentioned transformers like GPT-3 earlier, which have been developed for text and language processing, but we believe they can be powerful tools to analyze wireless signals.
On the video side, we are working with Lionel Oisel's lab on the next generation of video coding and compression tools. Specifically, the goal for the AI Lab is to develop a completely neural network-based codec for video.
Finally, the AI Lab is launching a new effort this year – again in sync with Jim Nolan's lab - around data services, to build a pipeline to ingest, process and use live (as opposed to synthetic or simulated) 5G data to train and optimize the neural network models we develop for wireless research.
CorpComm: One of your Lab’s major accomplishments was the introduction of CompareAI – the very first mobile app developed end-to-end in InterDigital’s Lab and the world’s first AI-enabled image compression app. In your words could you tell us about this pioneering app, what it revolutionizes, and who should download it?
Jean Bolot: We are very excited about the app. It is the very first iPhone app released by InterDigital and it is a great platform for users to experience our technology. In addition, it was very valuable for us to see the app’s development from scratch all the way to being published on the AppStore. Much thanks also to Olivier Le Quéré and the legal team who were critical in helping us navigate through the agreements and legal issues.
The app is unique because it is the first available implementation of a completely neural network-based compression for images. This is also the first time that these neural networks are enabled to run on a resource-limited device like a smartphone. This is important because most large neural networks run on big GPUs and CPUs in data centers and configuring them to work on commercial phones without draining the battery is still a challenge. Finally, the app lets users visually compare the quality of compressed images obtained with both neural network-based and with traditional techniques like JPEG or HEIC.
Next step for us is to release a deep video compression app later this year. But for now, please don’t be too harsh on these new AI-based techniques 12 This is still the beginning and just having them run on the phone is an achievement!
CorpComm: If you were to look into a crystal ball, where do you think AI and deep-learning algorithms will bring the greatest impact to our industries, and society at large?
Jean Bolot: Three decades ago, Mark Weiser coined the term "ubiquitous computing" to describe a world in which computers would be everywhere – in data centers but also door locks, watches, cars, smartphones of course and basically around our world today. I believe we need to prepare for "ubiquitous AI", namely a world in which all these computers have advanced AI capabilities. Ubiquitous AI raises many challenging issues of robustness, privacy, or energy use, but it also opens the door to boundless new opportunities. What AI can do now would have been considered magic 15 years ago. It's our turn now to invent the magic ahead.
CorpComm: Your department works tirelessly to make AI Algorithms smarter, and more predictive, to understand users and systems better. In an effort to better understand YOU, we’d like to offer a few random prompts for rapid word association, and why, to better get to know the man behind the AI.
First word is: HAL
Jean Bolot: ipHone + ALexa
CorpComm: Next is: Season?
Jean Bolot: 4 seasons. Vivaldi. My favorite composer.
CorpComm: Next word is: Samoa
Jean Bolot: Tropical island. Hawaii. Would love to be there now!
CorpComm: What do you think when you hear the word: Michael?
Jean Bolot: Nothing special pops up in English but in French, Michel, I think of Mont St Michel. Amazing place.
CorpComm: And lastly, what do you associate with the word: resolution?
Jean Bolot: New year resolution. Failure. Except if only one. And grit.
CorpComm: Thank you Jean and thank you for letting us get to know you and learn more about the incredible work in the AI lab. We’re looking forward to these new developments and will encourage everyone to download the newest CompareAI app here.