As artificial intelligence is touching almost every part of our lives, delivering us smart voice systems, driverless cars and customer service bots for banking and health care, it is time for government to learn about and act on the promise of AI and its potential perils. Yet despite the promises and dangers of AI, the U.S. government is behind the curve. “We are “flying blind in our conversations and decision-making related to AI,” according to researchers from Stanford, MIT and elsewhere.
AI is too important to ignore, too complex for simple policy solutions and too interrelated to address in siloes. It’s time for a national conversation. And no one is better positioned to start it than Congress, through the creation of a bipartisan National Commission on AI.
{mosads}The National Commission should have two goals: Protect America’s technological standing and increase economic growth and innovation. The commission should advance guiding principles of preserving social norms and protecting values like privacy and personal autonomy. It would hold hearings, conduct research, and make recommendations for industry, the Executive Branch and Congress.
While AI often refers to computer programs and software, here’s a more instructive definition: AI is an umbrella of modern technologies designed to mimic human behavior or thinking.
According to McKinsey’s Global Institute’s study of 800 occupations, fewer than 5 percent of jobs will be completely automated, and about 60 percent of such jobs have constituent parts (at least 30 percent) of their activities which could be automated.
This does not necessarily mean immediate or dramatic total job losses, but rather a dramatic set of changes in the nature of work. Even those job losses need to be compared to, or offset by, job changes and gains.
McKinsey sees AI as a “catalyst of future labor [that] could create demand for millions of jobs by 2030.” Among the new jobs are health-care providers, infrastructure developers, jobs for raising energy efficiency and others that don’t exist today. The biggest offsetting job gains will be from consumer spending freed up from cheaper AI-supported goods and services.
Automation and AI are changing the workflow at factories, hospitals, fast-food companies, and schools, making certain tasks irrelevant — but it’s not all dire news for workers. According to Gartner Research, AI will create 2.3 million new jobs by 2020. Some positions, like programming robots, are related to the new technology. And MIT Sloan Management Review predicts AI will create new careers.
Companies with AI systems will need human workers to “train” the technology and create its algorithms, for instance, and explain the complexity of AI inner-workings to non-tech co-workers. Finally, AI will need so-called “sustainers” — making sure the AI systems run correctly.
Other positions may be created through using AI as a collaborative tool. Gartner’s analyst Tina Nunno says “the real secret here is AI augmenting people.” Joe Jones, original founder of the Roomba — the automated vacuum, one of the first places customers began interacting with robots at home — also thinks robots will free up humans for higher-level tasks. AI can increase productivity. In turn that expansion elevates income which in turn means more money is spent on goods and services, boosting the economy.
He’s got a point: AI has equaled or sometimes surpassed the abilities of humans in performing certain tasks. Among them are, transcribing speech, spotting basic objects in images and defeating experts at highly complex games — including Go and poker.
Still, AI requires the support of human expertise in order to be used for most jobs. Reasoning, problem-solving, and strategizing are still beyond the grasp of machines.
That’s why we will need educated individuals to build, operate, and maintain these systems. Schools and programs that are strong in both creative and quantitative curriculum will need to step up to the challenge.
While AI degree programs at Stanford, Carnegie Mellon, and MIT will continue to produce in-demand engineers and developers, all K-12 schools — there are 37,000 in the US, according to the Department of Education — need a plan to integrate AI into their curricula to prepare students for future careers.
The private sector’s role in AI is also crucial. In health care, for example, the possibilities for improving care and lowering costs through AI are enormous but unrealized. About 30 percent of the stored data in the world is generated in health care, but only 180 of 6,000 US data scientists work in the industry. Still, novel studies demonstrate the promise of better care and lower costs through the application of AI to predict and reduce hospital falls, detect and prevent acquisition of hospital-borne infections and heart attack prevention.
A fully-integrated national AI policy must leverage the massive tens of billions of dollars federal and private sector investment in electronic health records (EHRs) and bring the promise of AI fully into day-to-day health care.
Software and algorithms designed by humans will likely never be free of human bias. In the past year, several incidents of AI systems demonstrating discrimination based on gender, race, and religion have hit headlines. AI developers must voluntarily learn and disclose conscious and unconscious biases and work to counteract them in the technology they build.
Efforts to assure diversity in people and datasets are an essential part of the social compact between companies and society. Addressing these issues is the only way our society will build and train AI technologies that offer social good for all.
Rarely do public officials have such a momentous challenge as they face in dealing with the public policy challenges of AI. This is an opportunity for bipartisanship (see the Future of AI Act). But it is also time for a uniquely American approach.
So far other nations have begun to sketch their plans, with China focusing on national competitive advantages and Canada focusing its policy lens on building the human and computing education, training and research infrastructure.
Now it’s time for Congress to seize the leadership role, addressing the social and values challenges and opportunities posed by AI. Sen. Maria Cantwell (D-Wash.) and others have begun the process, introducing legislation that addresses how policymakers should consider AI.
The most important thing government to do is to promote the social and economic benefits of AI through well informed policies. We can’t take it for granted.
Tom Daschle is former Majority Leader of the United State Senate and CEO of the Daschle Group; David Beier is Managing Director of Bay City Capital, a San Francisco-based venture capital firm.