李開復: 警惕人工智慧革命演變成科技「軍備競賽」| 金融時報專欄

本周，來自全球各地的行業領袖聚集在中國烏鎮，參加世界互聯網大會。加強人工智慧領域的全球合作，以及應對人工智慧給社會帶來的挑戰，是目前兩項重要的應時之需。

第一，因兼具多種結構性優勢，中國已迅速崛起為人工智慧領域的全球領袖。

這些優勢包括：龐大的高價值資料庫（人工智慧應用的關鍵）；實力漸長的企業和研究所輸出的大量頂尖人工智慧科學家；趨於完善且極度活躍的科技創新環境；以及通過國家扶持和強大資金後盾來推動人工智慧發展的切實意願。

我預期中美將共同引領人工智慧革命——這件事本身就是一個具有相當地緣政治重要性的新發展，在此過程中也需要其它國際參與者以開放的心態加入互動。

科技的迅速發展將令人工智慧在執行分析和量化任務時的表現超過人類。就DeepMind公司所研發的AlphaGo Zero而言，其成功只是一個開始。科技（深度學習和強化學習等）使之成為世界頂尖的圍棋選手，而今在金融服務、客服、廣告宣傳優化以及股票選擇等方面，類似科技的表現同樣遠超專業人士。即使是在高級任務的處理上，人工智慧也優於人類，如產品裝配、人臉和語音識別、無人機駕駛、乃至高速公路上的無人駕駛。

第二，這些精準的人工智慧工具在效率方面將遠高於人類。

這意味著在不遠的將來，大量的工作將不可避免地被替代。我相信人工智慧和自動化技術將在未來的10到15年之內替代現今50%的工作。

這對我們與科技之間的關係影響深遠，而人類在培養美德過程中所構建的工作倫理將隨之成為犧牲品。人工智慧顛覆了一個由來已久的信仰，即工作和我們使用工作收入的方式賦予了我們生活的意義。

在一個絕大部分的電子營銷、客服、出納、會計和金融分析都將被自動化的世界裡，我們如何保證受此影響的人們不會失業，也不會因此而感到抑鬱、無能？我們如何應對越來越不可逾越的貧富鴻溝？

如今的危險在於，中國作家郝景芳在其科幻短篇小說《北京摺疊》中構想的反烏托邦未來或將成為現實。在郝女士的故事中，中國被劃分成上、中、下三個階層。這種世界變為現實的可能性應當迫使決策者們對人工智慧可能帶來的後果給予認真的考量，並針對其帶來的挑戰尋求解決方案。

我們面臨著一系列難題。如何重新思考在工作中尋找尊嚴、成為社會中有貢獻的一員？如何重構教育系統，讓後代適應人工智慧的需求？如何實現財富的轉移，為落後於快速科技變化的人們提供機會？在人工智慧執行大量其他任務的同時，如何提高仍需由人類完成的服務和看護類工作的地位？

問題過於棘手，挑戰也過於廣泛，因此我們承受不起任由人工智慧革命演變成一場科技軍備競賽的代價。中美必須積極解決這些問題，而本周的烏鎮互聯網大會可以成為一個起點。

英文全文：

China and America must shape the high-tech future together

We need a global response to the huge job displacement AI is likely to cause.

Robotic arms at work in Guangzhou, China. Artificial intelligence undermines the age-old belief that it is work and its proceeds that give our lives meaning ? Bloomberg

This week, industry leaders from around the world gather for the World Internet Conference in Wuzhen, China. Two important developments point to the need for renewed efforts to enhance global co-operation on artificial intelligence and remedies to tackle the challenges it poses to society.

First, China has rapidly emerged as a global leader in AI with significant combined structural advantages. These include: a vast ocean of highly valuable data (which is critical for AI applications); increasingly strong companies and research labs turning out huge numbers of top AI scientists; a well-established and explosively ambitious technology start-up ecosystem; and genuine political will to advance the AI agenda with state support and massive financial backing.

I expect China will lead the AI revolution together with the US. This in itself is a new development of considerable geopolitical importance. It will require more open-minded interactions from other players on the world stage.

The rapid development of these technologies will enable AI to perform many analytical or quantitative tasks much better than humans. The success of DeepMind』s AlphaGo Zero was just the start. Now technology similar to that which enabled it to become the world』s best player of Go (deep learning, reinforcement learning and so on) are underwriting loans, providing customer service, optimising advertising campaigns and picking stocks significantly better than human professionals. Even for more advanced tasks, such as assembling products, recognising faces and speech, flying a drone, even driving on a highway, AI is now becoming more proficient than people.

Second, the fact that these accurate AI tools will be so much more effective than humans means that a massive job displacement in the near future will be inevitable. I believe that AI and automation will be capable of replacing 50 per cent of current job tasks in the next 10 to 15 years.

The implications for the way we live our lives with technology are enormous. One casualty of this profound shift will be a work ethic constructed around the cultivation of human virtue. AI undermines the age-old belief that it is work, and what we do with the proceeds of it, that gives our lives meaning.

In a world where most telemarketing, customer service, clerking, accountancy, financial analysis will be automated, how do we ensure that those bearing the brunt of these changes are not lost to unemployment, depression and feelings of worthlessness? How do we deal with the wealth gap becoming a chasm? The danger is that the dystopian future imagined by the Chinese writer Hao Jinfang in her science fiction short story 「Folding Beijing」 becomes a reality. In Ms Hao』s tale, the Chinese capital is divided into three different layers, for the upper, middle and lower classes.

The very idea that such a world might come to pass should concentrate the minds of policymakers grappling with the ramifications of AI and seeking solutions to the challenges it poses.

We face a series of profound questions. How can we reimagine what it means to find dignity in work and be a productive member of society? How do we redesign our education systems to meet the needs of future generations who will have to adapt to the demands of AI? How do we shift wealth and provide opportunity to those left behind by breakneck technological change? And how do we enhance the status of the service and caring jobs which will still need to be done by humans when AI is taking care of a range of other tasks?

These questions are so hard, and the challenges we face so vast, that we cannot afford for the AI revolution to turn into a technological arms race. The US and China will have to work together to solve these common problems. They can start in Wuzhen this week.