Delivered. accelerates AI with tech prodigy Alexandr Wang
Alexandr Wang – a tech prodigy and college dropout – has founded a $1-billion dollar Silicon Valley company at the forefront of the AI and machine learning revolution.
Developing new technology is like running on a treadmill notes Alexandr Wang, founder and CEO of Scale AI. “The trouble is, whenever you start to run faster, the treadmill speeds up,” he says. “You look forward and see there’s still so far to go.”
Scale AI is a three-year-old Silicon Valley-based company with a mission to accelerate the development of AI and machine-learning applications for its many customers. It made headlines because Wang, who dropped out of MIT to start it, is still only 22 years old; and the company has seen extraordinary growth since its founding in 2016. Yet many investors initially turned him down. “The vast majority were skeptical,” remembers Wang. “A lot of them thought it would be a niche business and weren’t sure how big it could get.”
But it got very big indeed. In August 2019, Scale AI was valued at over $1 billion, and Wang was included in Forbes’s 30 Under 30 in Enterprise Tech 2018. This sort of success could go to your head if you let it, he admits, but Wang has his eyes on a bigger prize: longevity. “The most important thing for me is to ensure that we’re always creating value for our customers, rather than anything vanity-related,” he says.
Those customers include automotive companies developing self-driving vehicles, financial services companies, healthcare firms and e-tailers. The one thing they all have in common is that they generate a lot of unstructured data, and want to use this data in a smarter way to be more effective or to develop innovative new products. Not all companies have the resources or the expertise to do this themselves, which is why they turn to Scale AI. Scale AI turns customers’ raw data into structured, labeled data to reliably train their AI applications. For example, Scale AI’s platform examines and labels millions of frames of data collected by vehicles rigged with sensors. That labeled data then trains the car’s software to recognize a wide range of objects and react appropriately to situations as and when they occur.
Naturally, this isn’t as straightforward as it sounds. “If you were teaching a child to recognize something – cars, for example – you would need to show them pictures of cars, and they would learn after a relatively small number of examples,” says Wang. “Machine-learning models, however, need an astronomical number of examples in order to perform at very high levels.” This problem is further compounded when it comes to teaching machine-learning models to predict what a pedestrian or another driver might do. But the results can be revolutionary. As Scale AI notes, thanks to machine learning, “computers can now recognize images and audio, translate languages, generate realistic text and beat humans at games.”
Wang is something of a tech prodigy who, prior to his time at MIT, began working for Silicon Valley question-and-answer website Quora as a teenage engineering lead. Starting his own billion-dollar business hadn’t featured in his game plan by this point, but it was clear to him that machine learning was the future. “I used to think of myself as a humble engineer,” says Wang. “Now I think of myself as someone who has the unique opportunity to help advance the technology and application of AI and machine learning across the world. I feel lucky to have had the opportunities to get here at such a young age.”
Can you give an example of how AI and machine learning help your customers?
Take an e-commerce fashion company that has thousands of items listed on its website in a range of sizes and colors. To make sure shoppers can find exactly what they’re looking for, the company may want to improve search relevance or make personalized recommendations for shoppers. Machine learning models can enable them to do just that.
Are you excited about the possibilities of AI and machine-learning technology?
Extremely. Research in AI and machine learning is progressing extremely quickly. Whenever there have been research advances in one area of science, they’ve been followed by technology that impacts our day-to-day lives. Take smart speakers, such as Alexa, which only exist because machine learning helps them recognize speech efficiently and effectively. So this is tech that gives us an incredible ability to improve people’s lives with incredible products.
Such as fully autonomous cars on our roads?
Well, I’m no better at predicting when we will see self-driving cars on the roads than anyone else; but I do believe the technology is improving every month and the end is in sight. It’s hard to say: “Oh, it will happen by year X.” But I feel confident that it will happen because we have scoped out the problems we need to solve. It’s now akin to the space race. Whenever we reach it, I think society will reap the benefits.
Are people right to be concerned about AI and machine learning?
I think there’s confusion – caused in part by the machine-learning community – about the extent of the technology. Machine learning can solve image recognition and speech recognition tasks and simple classification and prioritization tasks. But it can’t replicate the full extent of what humans can do. It will be a very long time before a human can have a conversation with an AI machine ... and not realize they are talking to an AI machine. — Tony Greenway
Published: November 2019
Image: Ian Tuttle for Delivered.