Jump to content

Scale AI

From Wikipedia, the free encyclopedia

Scale AI, Inc.
Company typePrivate
IndustryInformation technology
Founded2016; 8 years ago (2016)
FoundersAlexandr Wang
Lucy Guo
HeadquartersSan Francisco, California, U.S.
Key people
Alexandr Wang (CEO)
Number of employees
600 (2024)
SubsidiariesRemotasks, Outlier.ai
Websitescale.com

Scale AI (Scale) is an artificial intelligence (AI) company headquartered in San Francisco, California. The company provides labeled data used to train AI applications.

Background

[edit]

Scale was founded in 2016 by Alexandr Wang and Lucy Guo who had previously worked at Quora.[1]

When the company was first conceived, it supplied human labour to perform tasks that algorithms could not. Accel partner Dan Levine offered to provide $4.5 million in seed funding to Scale and his basement as a temporary headquarters. Within months, Wang and Guo realized Scale could fill the needs for autonomous vehicle (AV) companies by being able to review and label its driving footage data used to train their AI applications. Other investors of Scale included Dragoneer Investment Group, Tiger Global Management and Index Ventures.[2][3][4]

In 2018 Guo left Scale "due to differences in product vision and road map".[2]

AV-related companies are the main source of revenue for Scale. According to a June 2019 fundraising pitch deck seen by Forbes, Scale was on track to have an annual revenue over $40 million.[2][3]

In August 2019, after Peter Thiel’s Founders Fund made a $100 million investment in Scale, its valuation exceeded $1 billion and it acquired Unicorn status.[2][3]

By July 2021, Scale had reached a valuation of $7 billion. There was an increased demand for data labelling from clients in different industries.[4]

In January 2023, Scale laid off 20% of its workforce.[5]

In March 2024, Scale reached a valuation of almost $13 billion after Accel lead another round of funding.[6] In May 2024, Scale raised an additional $1 billion with new investors including Amazon and Meta Platforms. Its valuation reached $14 billion.[7]

In December 2024, Scale was sued by a former employee with allegations that it was committing wage theft and misclassifying workers.[8]

Military affiliation

[edit]

Scale also has been serving clients in the government such as the United States Armed Forces. Scale has pitched itself as a company that will assist the U.S. military in its existential battle with China by offering to pull better insights out of data, build better AVs and even create chatbots that can help advise military commanders during combat. Wang, who describes himself as a "China hawk," stated the U.S. will not be able to maintain its technological edge over the rising military power of China without AI developed by private technology companies. He was partially influenced by a trip to China in 2018 to see its AI scene and noted how the government there could use commercial technology for its own benefit.[1][9]

In January 2022, Scale was award a $249 million contract by the U.S. Department of Defense to Accelerate Government's AI Capabilities.[2][1][9]

In the fall of 2022, Scale planned to enter a deal with TikTok where it would provide insights for its advertisers. Despite opposition, Wang pushed for the deal to go through as the commercial opportunity was too good to turn down. However, the deal was eventually scrapped due to national security concerns.[9]

As of February 2024, Scale has made almost $80 million from government related contracts.[9]

Remotasks

[edit]

As the demand for data labeling from AV increased, Scale found it harder to keep up with the demand for human labour. Initially Wang turned to outsourcing agencies but the cost quickly increased. Gross margins were around 65% in early 2018 and had decreased to 30% by the fourth quarter of the year.[2]

In 2017, Scale established Remotasks as its in-house outsourcing agency. It set up over a dozen facilities in Southeast Asia and Africa to train thousands of data labelers. Remotasks has been established as a separate brand for client confidentiality according to Scale. Early employees say this was done to make Scale's strategy less obvious to competitors and shield the company from scrutiny.[2]

Remotasks was successful in controlling costs and by mid-2019, Scale's margins had recovered to 69%.[2]

According to a 2022 study by University of Oxford researchers, it concluded Remotasks met the "minimum standards of fair work" in just 2 of 10 criteria. It noted that the "obfuscation" of its association with Scale creates confusion that "can contribute to workers’ vulnerability to exploitation." The Lead researcher Kelle Howson compared Remotasks to garment factory workers in the same countries stating "There is pretty much zero accountability for those working conditions".[2][10]

In August 2023 The Washington Post released an article on Remotasks' presence in the Philippines. It stated that Scale paid workers at extremely low rates, routinely delayed or withheld payments and provided few channels for workers to seek recourse. Freelance review has to go through several layers of review and only after approval are payments credited to workers through platforms like PayPal. Payments can be withheld with no explanation and if work is rejected, the workers can be asked to redo the work with little to no payment at all.[10]

Remotasks pays workers on a task-to-task basis without any contract binding them to the company, which means workers often have few reliable ways to contact supervisors or escalate complaints. An employee has stated he was locked out of his account after raising a complaint to his supervisor about the number of tasks he'd completed without receiving payment.[10][11]

In March 2024, it was reported that Remotasks had terminated its services in whole countries without explanation. Workers in Kenya, Nigeria, and Pakistan found themselves cut off from accessing its services. Remotasks has also halted new sign-ups in Thailand, Vietnam, and Poland. A Remotasks representative claimed an administrative error was to blame for the lack of notice provided to contractors in Kenya.[11]

See also

[edit]

References

[edit]
  1. ^ a b c De Vynck, Gerrit (October 22, 2023). "Some tech leaders fear AI. ScaleAI is selling it to the military". The Washington Post. Archived from the original on March 12, 2024. Retrieved April 19, 2024.
  2. ^ a b c d e f g h i Cai, Kenrick (April 11, 2023). "How Alexandr Wang Turned An Army Of Clickworkers Into A $7.3 Billion AI Unicorn". Forbes. Archived from the original on April 6, 2024. Retrieved April 19, 2024.
  3. ^ a b c Vance, Ashlee (August 5, 2019). "Silicon Valley's Latest Unicorn Is Run by a 22-Year-Old". Bloomberg.com. Archived from the original on May 25, 2023. Retrieved April 19, 2024.
  4. ^ a b Loten, Angus (July 12, 2021). "Scale AI's Rapid Growth Reflects Widening Demand for Smart Software". WSJ. Archived from the original on February 26, 2024. Retrieved April 19, 2024.
  5. ^ Chatterjee, Poulomi (January 16, 2023). "Scale AI was Scaling Till Clienteles Refused to Buy it Anymore". Analytics India Magazine. Archived from the original on April 19, 2024. Retrieved April 19, 2024.
  6. ^ Clark, Kate (March 28, 2024). "Scale AI Nears $13 Billion Valuation in Accel-Led Round". The Information.
  7. ^ León, Riley de (May 21, 2024). "Amazon, Meta back Scale AI in $1 billion funding deal that values firm at $14 billion". CNBC. Archived from the original on May 23, 2024. Retrieved May 23, 2024.
  8. ^ Council, Stephen (December 13, 2024). "SF tech startup Scale AI, worth $13.8B, accused of widespread wage theft". SFGATE.
  9. ^ a b c d Jeans, David. "Defense Contractor Scale AI Quietly Scrapped Deal With Chinese-Owned TikTok Over Security Concerns". Forbes. Archived from the original on April 19, 2024. Retrieved April 19, 2024.
  10. ^ a b c Tan, Rebecca (August 28, 2023). "Behind the AI boom, an army of overseas workers in 'digital sweatshops'". The Washington Post. Archived from the original on May 24, 2024. Retrieved April 19, 2024.
  11. ^ a b Branndom, Russell (March 28, 2024). "Scale AI's Remotasks platform is dropping whole countries without explanation". Rest of World. Retrieved April 19, 2024.
[edit]