機器人産業群相與人類就業問題：報紙的說法

蘋論搞錯了某些面向，新型的工作會被開創。

蘋論：機器人 就業終結者
2016年07月27日

在商場和家裡工作的機器人時代已經來臨，他們不但端茶送水、幫忙提東西，還會在你想找人講話時陪你聊天。

現在和朋友聚餐加入的是外傭，很快你將看到一群聚餐親友的每個人背後站著一個可愛的機器人。貴婦們會競相比較誰的機器人是精品名牌，誰又換了新型號的名牌機器人。這是美麗新世界嗎？憂慮的是被機器人取代工作的勞工可能將永遠失業。

「新時代已來臨了」

鴻海與日本軟體銀行合作研發的服務型機器人Pepper前天正式在台亮相，獲得一銀、家樂福、國泰人壽、台新銀、亞太電信等企業首先使用。Pepper將以租賃方案切入台灣市場，含硬體、消費者互動服務、商業套件服務和售後服務等，月租費26888元台幣。這件事告訴我們一個新時代來臨了。
美國麻省理工學院（MIT）的《科技評論》自2001年開始，每年公布當年的「10大突破性技術」。此前公布的「2016年10大突破性技術」包括：免疫工程、精確編輯植物基因、語音介面、可回收火箭、知識分享型機器人、DNA應用商店、高效能太陽電池板、Slack通信軟體、特斯拉自動駕駛儀、Wi-Fi取電。其中知識分享型機器人的原型我們已經看到。
這種機器人可以學習任務，同時把知識傳到雲端，供其他機器人學習。其重要意義在於：如果不需要分別對所有類型的機器人單獨寫程式，就可以極大地加快機器人的發展進程。 

急需人工智慧人才

機器人的種類目前有二，工業機器人和服務機器人，前者早已用在工廠，形成自動化生產，像是汽車製造廠，已經排擠了許多工人的就業；後者是未來市場的寵兒，但也遠比工業型複雜，需要與人互動的能力，像是語音辨識、雲端運算、機器學習以及各種的感測技術等人工智慧。這個領域非常廣，還缺乏很多人才，年輕人要盡快學習這方面的技術，可以剛好趕上潮流，發揮所長。
日後還會出現機器人士兵，從事作戰任務、給需要異性或配偶的人機器人伴侶。令人擔憂的是以後櫃檯員、公寓管理員、保全、餐廳服務員、空姐、老師等工作都可能由機器人取代，這麼龐大人口的失業將是社會的災難。政府有義務辦理給予民眾新技術教育的機會，協助學有新技術人們就業。 

October 8, 2015 6:08 pm

The human cloud: A new world of work

Sarah O’Connor, Employment Correspondent

©Jacey

Nestled in his “man cave”, a room crammed with cardboard boxes and fishing lures in his Rhode Island home, Set Sar is earning money by letting a company track the tiniest movements of his eyeballs through his computer’s webcam.

About 10,000 miles away, Adi Nagara is hiding from the heat in his air-conditioned bedroom in Jakarta, researching an Indonesian industry for a consultancy firm. Though they are doing different tasks for wildly different sums on different sides of the world, the two men are connected. They are both members of the “human cloud”.

Some of these tasks are as simple as looking up phone numbers on the web, typing data into a spreadsheet or watching a video while a webcam tracks your eye movements. Others are as complex as writing a piece of code or completing a short-term consultancy project.Employers are starting to see the human cloud as a new way to get work done. White-collar jobs are chopped into hundreds of discrete projects or tasks, then scattered into a virtual “cloud” of willing workers who could be anywhere in the world, so long as they have an internet connection.

The uniting factor is that these are not jobs but tasks or projects, performed remotely and on-demand by people who are not employees but independent workers. Much of it is, in effect, white-collar piecework. Employers spent between $2.8bn and $3.7bn globally last year on payments to workers and the online platforms that act as intermediaries in the human cloud, according to a recent Staffing Industry Analysts report.

To its champions — the people who run platforms and others who believe we are on the threshold of a flexible work revolution — the human cloud promises to eliminate skill shortages, ease unemployment black spots and create a global meritocracy where workers are rewarded solely for their output, regardless of their location, education, gender or race. Some even say it could return us to the age of the cottage industry, before we crammed into factories or offices and lost control over our work.

“What we see today is people taking ownership again of the means of production, because you just need a computer, your brain and a wifi connection to work,” says Denis Pennel, managing director of Ciett, the international lobbying organisation for private employment agencies. “So actually, Marx should be very happy!”

Critics turn to history for their analogies too, but they talk of dead-eyed operatives on production lines, not happy artisans. In the human cloud they see a wild west of unregulated virtual sweatshops, breaking down service sector work into its constituent parts, making people compete in a worldwide race to the bottom. “It makes Adam Smith’s famous division of labour in pin-making look modest,” says Guy Standing, an academic and author of several books about the “precariat” and the growth of insecure work.

Turkers and nerds

Whether the human cloud is more utopia or dystopia depends, at least in part, on where exactly in its hierarchy you find yourself.

Mr Sar is near the bottom, as he readily admits. “We’re just getting crumbs as far as what we’re getting paid for it,” says the 29-year-old from Providence, capital of America’s smallest state.

He joined the human cloud through Amazon’s Mechanical Turk, a site run by the online retailer where “requesters” pay “Turkers” to do simple microtasks that humans are still marginally better at than computers, such as transcribing audio clips, filling in surveys or tagging photos with relevant keywords. The name Mechanical Turk refers to a fake chess-playing machine from the 18th century that fooled onlookers into believing it was an automaton when in fact there was a person hiding inside. Amazon — whose tagline for the platform is “artificial artificial intelligence” — calls the jobs on offer “human intelligence tasks”, or HITs. Many of them only pay a few cents apiece.

On a good day, Mr Sar would earn about $5 to $7 an hour by doing batches of HITs in his free time (he also has a job in a warehouse). But after Amazon increased the fee it charged “requesters” to post HITs to 20 per cent of what they pay workers, he says HITs dried up and pay rates dropped. “Now we as workers have to be competing against other workers to grab these good HITs.” Lately he has discovered a newer site, Sticky Crowd, which shows him videos and web pages and uses his webcam to track exactly what he looks at and what he ignores — useful information for advertisers. The pay is better: a dollar for every 2-3 minutes of eye-tracking.

Not all the work on offer is so futuristic. Take the cloud call centres that assemble armies of “independent agents” who work from home, pay for their own phone and internet, and only get paid when they are actually on a call. The average “talk-time rate” at one large cloud call-centre is $0.25 per minute, though some clients also offer sales commission.

Further up the hierarchy are platforms like Upwork, Freelancer and People per Hour, which feature more skilled tasks such as copywriting, IT and design work. Upwork, formed last year by a merger of two large platforms, is now the behemoth of the human cloud, processing about $1bn worth of payments from employers to workers last year (of which it takes a 10 per cent cut). The company took 10 years to reach $1bn, but reckons it will reach $10bn in another six. “It’s a thing that takes a long time at the beginning,” says Stephane Kasriel, its chief executive. “Then at some point it hits a tipping point, it becomes mainstream.”

Some of these sites invite workers to “bid” for the tasks on offer — specifying how quickly and for what fixed price they could do the work. Others offer payment by the hour. In most cases, employers and workers give each other star ratings after they finish a task, much like on eBay or Airbnb, allowing them to build a track record. Reputations are important: human cloud platforms know they need to link employers with good workers to encourage return visits, so many are starting to use “big data” algorithms to recommend certain workers for certain gigs.

People per Hour has set up a sister site, SuperTasker, which uses a smaller group of pre-screened workers to do fixed tasks in a fixed period of time for a fixed price: a 400-word blog post delivered in three hours costs $45, for example. Xenios Thrasyvoulou, the company’s founder, calls this “SKUs for work”. SKUs are stock keeping units — retailer shorthand for indistinguishable products.

Yet at the top of the human cloud’s hierarchy, “standardisation” is a dirty word. “It’s not a commodity — clients don’t choose on price,” says Daniel Callaghan, chief executive of UK-based MBA & Company. His platform, like US rival HourlyNerd, links companies with “consultants on demand”. Other specialist sites include Topcoder for computer programmers and Upcounsel for lawyers.

Consultants on MBA’s platform charge between £250 and £4,800 a day (then MBA adds its fee of 20 per cent). Mr Nagara, who is 30, is one of the platform’s consultants; he used to work for Australian investment bank Macquarie but moved back to Indonesia for family reasons. He reckons he earns more from his daily rate than he would as a full-time employee — though he sacrifices the security and benefits, a trade-off his parents do not understand. “They stayed with one company for decades, so when they see me being unemployed every two months they think ‘Jesus!’ They probably think I’m a disaster!”

Opportunities and costs

While many workers on these specialist sites are young and fleeing the corporate grind or topping up incomes, others are capitalising on a lifetime’s worth of knowledge. Nasa, the US space agency, once posted a challenge to find an algorithm that could predict solar flares: the winner (who Nasa paid $20,000) was a retired radio frequency engineer.

Podcast

Brazil: Dilma Rousseff in the line of fire

Latin America’s biggest economy is in recession, the political elite is engulfed in a huge corporate scandal and the president faces impeachment in a move that could bring the country to a standstill, reports Joe Leahy

It is not hard to see the promise of the human cloud for employers, who frequently complain about skills shortages and a lack of skilled migrant workers.

Mr Callaghan says the human cloud will make such problems disappear. “You can now get whoever you want, whenever you want, exactly how you want it,” he says. “And because they’re not employees you don’t have to deal with employment hassles and regulations.”

That is particularly useful for fast-growing start-ups. Dom Bracher, a 22-year-old founder of UK-based mobile marketing company Tapdaq, uses developers and designers in Scandinavia and central Europe. “There’s no need for someone to be in the same city as you,” he says.

Susan Lund, a partner at the McKinsey Global Institute, says the human cloud can improve social mobility, too, since it allows people to amass hard evidence of their abilities regardless of the formal qualifications on their CVs.

“For somebody who doesn’t have a degree from a top university or even a degree at all, accumulating those ratings is very important — to be able to say I’ve done X hours of coding and my average rating was Y — that’s very powerful,” she says.

The other consequence of work moving online is that more people should be able to do it: the housebound or people in locations where job opportunities are scarce.

The flipside is that workers in places where the cost of living is lower can undercut their peers in more expensive countries. “You can have someone in Gothenburg competing against someone in Dakar,” Prof Standing says.

Plenty of IT and call-centre work has been outsourced to countries like India but Prof Standing believes the next wave of “silent offshoring” will be more devastating for wages and conditions in the developed world.

It is hard to test this hypothesis, since most human cloud platforms are not listed and only disclose their data selectively. Still, a lot of work appears to gravitate to low-cost countries with skilled workforces: Upwork’s biggest markets after the US by worker earnings are India, the Philippines, Ukraine and Pakistan.

But Mr Sar and Mr Nagara are evidence that the picture is complex: low-paid work does not always drift east and high-paid work does not always drift west.

Contractors or staff?

Perhaps the thorniest problem of all for the human cloud is one that has also plagued Uber, the taxi app: when should an independent worker actually be classed as an employee?

Human cloud platforms usually classify workers as self-employed, which frees them from the requirement to pay minimum wages, employer taxes and benefits like sick pay.

But lawyers and workers are challenging them: last year a cloud platform called Crowdflower offered more than $500,000 to settle a US class action lawsuit from workers who said they were really “employees” and were therefore owed the minimum wage.

Most countries’ legal systems are struggling to keep up with these new forms of work. “In these arrangements, there’s really more than one employer — the law can’t grapple with this,” says Jeremias Prassl, a law professor at Oxford university.

Jonas Prising, chief executive of ManpowerGroup, an employment agency, predicts policymakers will impose more regulations on the new platforms soon.

“Who is taking care of these individuals? Who is providing the security in terms of taxation and social security? Who is doing the work is not known, who is paying the tax is not known, the age of the people doing the work is not known,” he says.

For all that, it can be a false comparison to contrast “insecure” human cloud work with “secure” traditional jobs — particularly at the bottom of the economic ladder.

Mr Sar has a job in a warehouse, but like many low-paid employees in developed countries, his rights and protections have been hollowed out. He is employed arm’s-length by an agency, which means he can be fired on the spot and is ineligible for many benefits. In the warehouse he wears an earpiece called “The Jennifer unit”, a robot in his ear that tells him what to do and tracks his performance and his downtime.

The human cloud might not pay much, it might be monotonous, but it gives him a sense of control. “Growing up through the years I’ve always worked for someone else. You’re treated as a number and not a human,” he says.

But his work in the cloud is different. “I can stop whenever I want. I can take a break, or eat something,” he says. “The idea of being my own boss is what really attracted me.”

作者：日經中文 2015-09-28 日經中文

日本的機器人産業正在蓬勃發展。日本軟銀集團面向一般消費者推出人形機器人「Pepper」首批1000台在1分鐘內即告售罄。無人機被廣泛應用於各種拍攝活動。此外，機器人技術也在自動駕駛汽車及電子商務的基礎設施方面提供支持。追溯這些技術發展的源頭，必然會碰到一名日本研究人員。



他就是索尼電腦科學研究所的所長北野宏明（54歲）。

2015年6月下旬，在日本東京某餐廳內，北野笑著説：「如今機器人已滲透到社會生活中，形成了我們所期待的商業模式」。1998年，作為日本科學技術振興機構的創造科學技術推進事業（ERATO）的一環，「北野共生系統項目」啟動。加入該項目的技術人員在之後紛紛開創了新事業，他們被叫做「ERATO Founders（創業者們）」。

北野是日本人工智慧第一人，曾參加過索尼寵物型機器人「AIBO」的研發。在當時還僅是重視精密地機械運轉的機器人研發領域，他便提出了機器人獨立思考、與人類共存的理念。

作為北野項目的成果，人形機器人「PINO」曾在日本歌手宇多田光的音樂錄影帶中出現。這種採用市售零部件，造價僅需50萬日元（約合台幣13萬8050元）左右的設計成為商品賣點。當時負責銷售的是日本ZMP的社長谷口恒（51歲）。

谷口曾在大型控制機器公司任職，後成立了經營電子書的內容提供商企業。透過擔任北野項目的技術顧問，谷口與北野相識，對於與社會共存的機器人這一想法深受啟發。現在，谷口的公司已成為汽車自動駕駛技術的世界領跑者，並獲得了美國英特爾公司的投資，事業正步入發展軌道。

曾用於「PINO」製造的技術如今已遍地開花。進軍土木工程領域的iXs Research公司代表山崎文敬（40歲）研發高速公路的橋樑檢測機器人。擔任技術控制的是37歲的遠藤謙。他與原田徑運動員為末大等人共同創立了Xiborg公司，研發用於體育競技的機器人假肢。

機器人公司RT的代表中川友紀子（44歲）曾在北野的帶領下從事畫像識別技術的應用工作，研發出全球首個採用「安卓系統」的雙腳步行機器人。同時，中川也參與了法國阿爾德巴蘭公司開發的人形機器人「NAO」的銷售工作。阿爾得巴蘭也是人形機器人「pepper」的研發公司。著名機器人領域投資人、原谷歌高層安迪・羅賓（Andy Rubin）也是中川的朋友。








日本Flower Robotics公司社長松井龍哉（46歲）師從建築師丹下健三，作為設計師參與了「PINO」的研發。他希望設計出「可以融入家庭的機器人」。Flower Robotics是2015年7月下旬在中國舉辦的「RoboCup機器人世界盃賽」的贊助商，開始表彰設計精良的機器人。

機器人世界盃賽由北野倡議發起，以「機器人足球隊戰勝人類的世界盃冠軍球隊」為目標。

作為「pepper」的原型，「NAO」在機器人世界盃賽中積累了技術經驗。美國Kiva Systems公司被亞馬遜（amazon.com）收購後，為其物流倉庫提供機器人技術支持。Kiva Systems創始人拉菲羅·安德烈（Raffaello D'Andrea）也認為：「NAO在比賽中積累了技術」，該公司是目前全球最先進的無人機製造商之一。

此次被稱為日本的第3次機器人熱潮。第1次是上世紀70年代的工業機器人。進入21世紀後的寵物型機器人「AIBO」引發了第2次熱潮。第3次正值當下。受到北野的影響、個性迥異的創業家們正在為新型機器人社會的到來貢獻著力量。

【更多日經中文】

1.韓國愛茉莉挑戰化粧品亞洲第一

2.日本宅女愛「痛包」

3.「蒼蠅」也能成為「怪物」