Meet Andrew Ng, a 2022 datanami person to watch


Andrew Ng is one of the most influential people in the field of Big Data and AI. He’s also one of the busiest, with stints at Google and Baidu, not to mention the co-founder of Coursera and his latest ventures, Landing AI and DeepLearning.AI. We would be delighted if he added another reference to his excellent CV: datanami Person to watch for 2022.

Ng kindly answered our questionnaire, which follows.

Datanami: You’ve had a storied career, from your work at Google and Baidu to founding Coursera and now Landing AI. To what do you attribute your great success?

Andrew Ng: I find it difficult to answer, because I don’t think I’m having that much success! But looking back, two things that had a big impact:

I was ready to stick to a contrary position in which I believed.

In terms of personal productivity, I think a lot of people underestimate the importance of forming useful habits – like learning a little bit each week – so that your habits can get you going without having to constantly rely on willpower to get you going. things.

For example, developing deep learning algorithms (Google Brain), developing AI in China (Baidu), trying to provide high-level university education to millions of people (Coursera) or developing AI for manufacturing (Landing AI), all of these sounded like “bad” ideas to some when I started. But many of the most impactful projects result from spotting an opportunity that others haven’t yet spotted and the tendency to move in a new direction.

Having good teammates is also a big part of success.

Datanami: Can you tell us more about Landing AI and LandingLens?

Ng: LandingLens, Landing AI’s flagship product, is a tool that facilitates the creation and deployment of state-of-the-art deep learning algorithms for vision algorithms. Our initial focus is manufacturing and industrial automation.

Many factories rely on visual inspection to ensure that products are free from defects. Modern deep learning has greatly expanded the set of defects that can now be detected with computer vision, but is not yet widely used. Why is it? The two main barriers to adoption are:

  • Manufacturing datasets are small and most deep learning algorithms have been developed for much larger datasets.
  • Each factory manufactures a unique product and therefore needs a custom model trained to detect its defects. Unfortunately, there isn’t enough AI talent to create the number of custom models the industry needs. LandingLens is an MLOps (Machine Learning Operations) platform that allows manufacturers to use data-centric AI development – that is, to systematically design the data provided to the landing algorithm. training – to train a custom model that achieves high accuracy, often even when only a little data is available. We find that adopting Landing AI’s data-centric AI approach also reduces development time and deployment time.

Datanami: Large Transformer networks like BERT have been instrumental in bringing powerful NLP to the masses. Do you think this has brought us closer to Artificial General Intelligence (AGI)?

Ng: I find the idea of ​​AGI exciting, but it’s also very hyped. While BERT and other transformative neural networks (such as GPT-3) are a tremendous step forward for app makers, and have also demonstrated an ability to generate language across a surprisingly wide range of topics and styles, they are really just tiny steps towards AGI.

Building a ladder was a step toward getting humans to the moon, not because we got there by building a 239,000 mile ladder, but because we couldn’t have built rockets if we didn’t. had no ladder. Transformer models (like BERT) also seem like a big step, but I think the path to the AGI dream will take many more decades – if not centuries – of basic research and breakthroughs.

Datanami: What do you hope to see from the Big Data community in the coming year?

Ng: Big Data has been instrumental in the development of AI, and I hope all of us in Big Data will embrace and contribute to the data-centric AI movement.

AI systems are built using code (which implements a learning algorithm) and data (used to train the system). For many years, the conventional approach in AI was to download a dataset and work on the code. Thanks to this development paradigm, for many applications today, the code aspect of an AI system is largely a solved problem. You can download a template from GitHub that works quite well for your project. Rather than spending time on the algorithm, in many cases it is now more useful to work on the data, systematically iterating over the dataset using data engineering processes and principles .

Data-centric AI is an emerging technology approach that develops principles and tools to systematically design the data needed to build a successful AI application. I started talking about data-centric AI in a Youtube video on March 24, 2021 and since then I have noticed the phrase appearing on more and more corporate websites. I hope these companies, and many others, will create tools that may allow many people to apply these ideas more systematically.

This will be essential to democratize access to AI systems, as well as to open up many new applications where the amount of data available is not large.

Datanami: Outside of the professional sphere, what can you share about yourself that your colleagues might be surprised to learn? Unique hobbies or stories?

Ng: Most people don’t know that I love the arts. To grow, my father was a doctor who taught me science, computer programming and also AI. And my mother organized art festivals and frequently took my brother and I to music concerts, opera and theatre. So my upbringing had a good dose of arts and sciences.

Today my own artistic skills are limited to hand drawing the occasional panda for my daughter (who loves them) or playing the melody “BabyShark” on the piano for my son. But I always appreciate the creativity of artists who create pieces or performances that move the human spirit.

You can read the rest of our interviews with the 2022 datanami Person to watch program on this link.

Previous RBI could raise policy rates by 75 basis points cumulatively in the next 2 reviews: SBI Research | Economy News
Next Foodstock 2022: Concert, beer and food festival this summer