The Verge - Artificial Intelligences 10月16日 22:07
Uber 启动微任务试点,让司机和快递员为 AI 训练赚取额外收入
index_new5.html
../../../zaker_core/zaker_tpl_static/wap/tpl_guoji1.html

 

Uber 正在其美国市场推出一项新的试点计划,允许其司机和快递员通过完成“微任务”来训练人工智能模型,从而赚取额外收入。这些任务包括录制语音、上传图片和提交特定语言的文档,例如上传西班牙语菜单可能获得高达一美元的报酬。此举旨在利用 Uber 全球庞大的自由职业者网络,为 AI 数据标注领域注入新的活力,并与 Scale AI 和 Amazon Mechanical Turks 等现有平台竞争。此项计划是 Uber 致力于打造“灵活工作最佳平台”战略的一部分,同时还包括改进行程报价卡、推出高需求区域热力图以及增强女性司机对行程的控制权等功能。

🚗 **AI 数据标注新机遇:** Uber 推出微任务试点,让美国司机和快递员能通过完成诸如语音录制、图像捕捉和文档提交等简单任务来训练 AI 模型,从而赚取额外收入。这为 AI 训练数据提供了新的、低成本的来源,并可能挑战现有的数据标注服务商。

💡 **灵活工作平台升级:** 此举是 Uber 致力于打造“灵活工作最佳平台”愿景的一部分。除了微任务,Uber 还将改进行程报价卡,提供更多信息和决策时间;推出高需求区域热力图,帮助司机更高效地接单;并为女性司机提供更多控制权,如仅接女性乘客的选项,以及允许设置最低乘客评分。

⚖️ **公平性与保障措施:** Uber 正在改进其公平性措施,包括限制某些特定类型收入机会的访问,而非完全停用平台,以应对可能出现的问题。此外,在做出停用决定前,将允许司机陈述其观点,并对虚假投诉的乘客进行处罚。新的延迟行程保障和加强的提示,也旨在提高司机收入和乘客满意度。

In its quest to become the ultimate app for “flexible work,” Uber launched today a new pilot to allow its US drivers and couriers to earn extra money by performing “microtasks” to train AI models.

These tasks include audio voice recording, capturing and uploading images, and submitting documents in certain languages. The prompts will vary, but some examples include “upload images of cars” or “record yourself speaking in your language or local dialect.” Another example is to upload a menu that’s written in Spanish, which could earn someone as much as a whole dollar.

The move could position Uber’s global army of freelance drivers and delivery workers to challenge established players like Scale AI and Amazon’s Mechanical Turks — businesses and platforms that work with generative AI companies to help train their models by having humans annotate and label the data that feeds them. Most of that work is done through low-cost labor outside the US and is seen as critical in the development of powerful AI models.

It’s also one of many announcements that Uber is making today around the idea of building “the best platform for flexible work,” as CEO Dara Khosrowshahi declared today at an event in Washington, DC. Most of the announcements center on how drivers and couriers interact with the app, aiming to make the experience of working for Uber easier, smarter, and fairer. In addition to digital microtasks, Uber is also redesigning its trip offer cards, introducing a new heatmap to direct drivers to high-demand areas, and expanding its features to give women drivers more control over their rides.

Uber has used independent contractors to train AI models in the past, using what it calls “human-in-the-loop” processes that blend “human expertise with machine automation,” according to the company’s AI Solutions Group. The company recently bought Belgian startup Segments.ai to expand its nascent data-labeling business. Uber drivers in India were recently given the opportunity to earn small amounts of money responding to prompts in the Uber app for the first time. Now the company is bringing its digital tasks pilot to the US.

It remains to be seen whether Uber drivers will take to these microtasks, given how many of them already complain of low pay as a result of the company’s high take rate on rides and deliveries. Of course, Uber classifies drivers as independent contractors, arguing they are in business for themselves and thus ineligible for traditional benefits like overtime, minimum wage protections, and health insurance. Some Uber drivers argue that the company’s algorithm exerts far too much control over their lives to be viewed as anything other than employment.

In addition to digital microtasks, Uber is also changing the offer cards for drivers and couriers to give them more time and information before accepting the trip. Offer cards are what a driver sees before they accept (or reject) a trip request. Now Uber is giving them more time to make a decision when the card first appears in their app. Uber is also rolling out a new on-trip experience for couriers that “simplifies” multi-order deliveries with clearer pick-up and drop-off details, as well as alerts for commonly missed items.

A new heatmap has been designed to give drivers more transparency about high-demand areas. Red areas indicate the shortest wait times, followed by orange, then yellow. Purple areas are where surge pricing is in effect, and it shows how long drivers waited on average for a trip based on recent data. And drivers who are commuting from their homes to a high-demand area, like an urban center, can now choose between a route that gets them there as fast as possible or one that maximizes fares along the way.

Uber is also making some changes it says will make its app safer for both riders and drivers. After first launching its Women Rider Preferences feature in the US last July, the company is now making it available in more cities, including Baltimore, Minneapolis, Philadelphia, Seattle, Portland, and Washington, DC. This feature allows women drivers to set their preferences so they only receive women riders, and vice versa.

In markets where the feature is available, Uber says that women have used it on over 100 million trips. A quarter of women drivers turn it on every week, and more than half keep it on for more than 90 percent of their trips.

Uber is now letting drivers set minimum ratings for riders based on their comfort levels, toggling the feature on or off depending on the situation — for example, using it late at night and relaxing it during the day. Combined with consumer verification for riders, Uber says these tools help “give drivers more peace of mind on every trip.”

And lastly, Uber is rolling out several new updates related to fairness. Drivers have long complained of feeling helpless in the face of indiscriminate deactivations from the app and the laborious process to undo those decisions. Uber now says it hears those complaints and will strive to ease them — but not totally eliminate the need for deactivation:

We know that losing access to Uber can create real challenges, which is why we’re working to make it easier for drivers and couriers to keep earning, even if issues arise. Where we can, we’ll limit access to only specific types of earnings opportunities rather than the full platform.  For example, if an alcohol delivery issue is reported, drivers can still take food or rideshare trips. As always, serious violations, including safety issues, may result in losing access to Uber altogether.

If Uber receives a complaint from a rider about a driver, it says it will also allow the driver to tell their side of the story before making a decision. And if a rider files a false reports about a driver, that rider will risk deactivation.

Uber is also rolling out a new Delayed Ride Guarantee, in which drivers can earn more if a trip is delayed due to a customer or for extenuating circumstances, like traffic or detours. It is also beefing up its tipping reminders for riders, expanding them to iPhone Live Activities, so riders receive more nudges about tipping their drivers or couriers.

Fish AI Reader

Fish AI Reader

AI辅助创作,多种专业模板,深度分析,高质量内容生成。从观点提取到深度思考,FishAI为您提供全方位的创作支持。新版本引入自定义参数,让您的创作更加个性化和精准。

FishAI

FishAI

鱼阅,AI 时代的下一个智能信息助手,助你摆脱信息焦虑

联系邮箱 441953276@qq.com

相关标签

Uber AI Flexible Work Gig Economy Microtasks Data Annotation Driver App Artificial Intelligence Freelance
相关文章