The Democratic co-chair of the Congressional AI caucus has called on Washington to develop a long-term plan dealing with the issues surrounding artificial intelligence, and the repercussions to society of its introduction.
Speaking at Nvidia GTC in San Jose on Tuesday, the representative for California's 9th district said AI is going to be one of the main tools of international competition in the future, and the United States needs to step up.
"Whether it is developing technology, whether it is in education, whether it is making our cities liveable and safe, I think this is a real challenge to our nation to make sure that we do develop a long-term plan," he said.
"What I'd like to see is a five-year plan, or something of that nature, where we lay out our educational goals, we have a workforce development issue, we need to make sure that young people have opportunity, we want to make sure young people see tech and artificial intelligence as something that they want to do, and then we need to give them the tools so that they can become proficient, and then we need to actually put money into research and development so that we support public-private partnerships, so that some of the companies that are doing AI, big and small, can take advantage of some government-funded research programs."
Last month, the executive branch made its feelings known when President Trump signed an executive order prioritising AI development, and seeking to ensure American leadership in artificial intelligence.
To McNerney, the executive order fell short. The Congressman said it would be up to Congress to "pick up the slack", and that he would need to work with Republican co-chair Pete Olsen to "move policy ideas forward and get the Congress behind us".
On the issue of handling bias with AI systems, McNerney said it is important to work out if particular groups are being hurt by it, or advantaging others unfairly.
"When you create algorithms that impact people's lives, then we have to be a lot more careful, and there's a lot of awareness now about how these kinds of algorithms have been used in terms of getting a loan for your home, or in terms of what happens when you get stopped by a police officer," he said.
"Another case where bias is extremely important is in military applications, if you have a weapons system that's operating with an algorithm, it could have lethal consequences and so bias is a matter of life and death in those situations.
"It's a real challenge, and it's not a mathematical challenge, it's a social challenge."
With a doctorate in mathematics, McNerney would know. The Congressman said algorithm testing needs to be developed, and added that biased data is also critical to identify.
"If you have data that's badly bia,sed then you are going to have similar results, and how to determine that data is biased? Not only that, but getting your hands on good data is also a very big challenge," he said.
"The challenges are out there, and I think algorithmic bias and bias in AI applications is something that is going to require a good partnership between public and private entities, so another place I look forward to working with industry."
Disclosure: Chris Duckett travelled to GTC in San Jose as a guest of Nvidia
More from GTC 2019
- The future of graphics is unquestionably AI image generation: Jensen Huang
- Pascal GeForce GTX cards could get ray tracing with driver update
- Nvidia goes Nano for latest Jetson release
- Nvidia GauGAN takes rough sketches and creates 'photo-realistic' landscape images
- Nvidia unwraps RTX and T4-based hardware and cloud instances
- Pure Storage, Nvidia, Cisco launch AI hyperscale systems