Powering the World’s Progress in Artificial Intelligence

Gone are the days when the dream of AI seemed out of reach and reserved for Hollywood movies. Decades ago, AI was framed as a mythical technology that over-promised and never delivered. But today, AI is being developed and deployed around the world and is advancing science, healthcare and business at a blinding rate.

I sat down for a fireside chat with Dr. Ian Buck, NVIDIA's Vice President and General Manager of Accelerated Computing, for FedScoop's FedTalks to discuss NVIDIA’s role in fueling the advancement of AI and how the company is helping accelerate federal AI adoption.

Ian started with Alex Krizhevsky, who helped launch the AI revolution when he developed the first AI framework during an ImageNet competition in 2012, where his GPU-powered code unlocked AI's image recognition potential.

A lot has happened since then, and NVIDIA technology has emerged as the leading platform for AI computing. Ian explained that AI requires massive amounts of data and tremendous computing capabilities, which is a challenge for traditional CPUs. GPUs are often hundreds of times faster than CPUs for AI, allowing them to quickly train AI models using massive datasets, as well as instantly derive answers in production AI systems.

With its increased accuracy and superhuman results, AI has given the world a powerful new tool to solve challenges. AI is enabling self-driving cars, powering smart assistants like Alexa and Siri and is even impacting our purchase decisions with product recommendation services. Conversational AI allows computers to interpret, understand, answer and repeat speech. AI is rapidly modernizing medical imaging -- algorithms to diagnose diabetes and strokes have been cleared by the FDA for the first time.

So how can the federal government apply AI capabilities to solve its problems? While the challenge of AI development can seem daunting, government agencies need to just get started.

"My first advice is to do something,” Ian said. “You don't have to solve your biggest problems—AI certainly has amazing potential to do amazing things—but you learn a lot by doing something simple, easy and quick to dip your toe in the water and understand some of these new workflows."

Some easy places to start: data scientists in federal agencies can experiment with applying AI to their data science problems. NVIDIA offers trained neural network models and tutorials to help federal data scientists experiment with AI use cases like image recognition, speech synthesis, natural language processing, sentiment analysis and search applications.

Agencies can also take advantage of NVIDIA's Deep Learning Institute, which offers hands-on training in AI and data science.

Hear more about the federal government's AI journey from my chat with Dr. Buck: https://www.fedscoop.com/events/fedtalks2020/live/