How will artificial intelligence change the world? Many experts have spoken about how AI’s negative effects, from racial bias to job replacement to ecological costs, could usher in a grim future if left unchecked.
With rare insight into the companies developing cutting-edge AI and their plans for its use, AI expert Kai-Fu Lee has his own sobering predictions about the future of war, work, healthcare and more. But while we must contend with serious externalities, Lee is optimistic about our future. He believes that too much negativism could cause a downward spiral and make bad outcomes a self-fulfilling prophecy.
To present a more balanced view of AI technologies, Lee partnered with novelist Chen Qiufan to write AI 2041: Ten Visions For Our Future. Grounded in Lee’s expertise, the work of scientific fiction presents different visions of how AI could shape our lives 20 years from now across different industries and geographies. To share predictions and insights into our collective future, Kai-Fu Lee joined moderator John Markoff at a virtual CHM Live event on October 7, 2021.
Many conversations have focused on mitigating AI’s hidden or unintended negative effects, but how do we deal with AI that is designed to harm people? Kai-Fu Lee is deeply concerned about advances in autonomous weapons. Decades ago only large countries had the resources to develop nuclear weapons, but today the lower cost of autonomous weapons could allow many more countries to participate in an arms race. The first group of autonomous weapons will likely be drones because they are inexpensive and do not need to follow the rules of the road.
It is probably too late to completely ban these weapons or to enforce a model in which only a human can trigger weapons. Lee suggests built-in hardware that detects negative use and legislation to regulate these weapons. He hopes AI 2041 will urge experts to discuss global standards and limitations on autonomous weapons, though he fears it may take a catastrophe to shock the world into action.
According to Lee, fears that robots will take our jobs may not be unfounded. AI is ideally suited to take over routine tasks, which make up a large percentage of lower-wage jobs. Eighty percent of the AI companies that Lee’s Sinovation Ventures invests in aim to replace people as their primary value proposition. UiPath, a robotic process automation company, is figuring out how to replace 40-70% of entry-level jobs in areas like accounting and human resources.
Lee predicts that displacement of blue- and white-collar jobs will reach a peak as AI matures within the next five to 15 years. He fears “huge wealth inequality.”While those at the top levels of companies that replace labor with AI will become billionaires, those who lose their jobs may face a training crisis. Workers would learn how to do less-routine jobs, only to see them replaced again by advances in AI. This cycle of displacement and re-training could lead to depression, substance abuse, and even addiction, to a retreat into virtual reality or games.
AI will also create new jobs in areas like data labelling, robotic repair, and AI engineering, but Lee anticipates that job growth here will happen more slowly than job replacement. Following this trajectory, it will take 20-25 years for unemployment to level off and 30-40 years for many more interesting jobs to be created, making for a difficult transition period.
With more health data than ever becoming digitized, Lee believes we have an opportunity to reinvent the future of healthcare. Yet he sees serious conflicts between the AI and medical communities. For example, when an AI scientist creates a new cancer-detection algorithm to replace radiologists, the hospital might have concerns about their liability if something goes wrong and whether the budget can support it.
Another conflict centers around aging. While the mainstream medical community does not regard aging as a treatable condition, says Lee, the AI community has begun to think differently. By collecting more health data through bloodwork, wearable computing, and full-body MRIs, AI can determine a person’s relative health age rather than number of years. A doctor can then interpret those results and give recommendations to the patient. Lee himself is a “lab rat” for the companies providing these services and finds that making health data-centric helps motivate him to compete against himself. After meeting his quarterly health goals, which he compares to KPIs or OKRs, for a year, he has reduced his “age” from 59 to 53. He sees this as an example of how AI can make a huge difference in healthcare.
When will we have a robot assistant besides the Roomba in our homes? Lee says there first must be a reduction in the cost of robotics technology, but there are potential applications in the home. For example, John Markoff sees a need for a machine that can safely give an aging human a shower. One company has developed robots with soft, bubble-like fingers that can grasp fragile objects (like egg yolks) without breaking them. Though created for laboratories and factories, it’s possible that this technology could be adapted for the need Markoff described.
While elderly care by robots seems technically achievable, Lee wonders whether it is something people really want. He tells the story of a care robot installed in an elderly home. Though it could help people call for help, access entertainment, and purchase items, its most-used feature was customer service. This was not because people had issues with the product, Lee says, but because they just wanted someone to talk to. He thinks that when it comes to care, people would rather have family and friends, or at least another human, take care of them.
We face two possible future worlds, Lee says. In one, advances in technology will take us away from human interaction. As AI becomes part of every sector from education to entertainment, and as virtual reality becomes more realistic and compelling, there may be less opportunity and even less desire to interact with other people.
But there is another possible world in which technology pushes us to place greater value on human interaction. As AI takes over routine work, maybe we will have more time to spend with the people we care about. AI is not strategic or creative, and it cannot love. While some people may be okay with having an AI companion, most people will still want true human-to-human contact.
What do YOU think the world will look like in 2041?