In the wake of Spirit Airlines filing bankruptcy last May, tech giant Google has swept in to
claim the failed company’s data with a $10 million cash bid. Google outbid a $7.5 million
offer from AI data company Mercor for the massive data stash; however, the U.S.
Bankruptcy Court delayed the approval hearing for the sale until September 9.
The adjournment of the hearing, which was originally scheduled for August 19, is in lieu of
an objection filed by the Association of Flight Attendants (AFA), which represents the Spirit
crew.
The AFA said it is seeking restrictions on the sale of flight attendant employee data and
further protection for employees if the sale is approved.
Included in the huge data block are, reportedly, 100 million employee emails, 500 million
Microsoft Teams messages, spreadsheets, calendars, as well as marketing, productivity
and operations data—in all, approximately 30 million lines of code.
Spirit has said the data will be de-identified by a third party before being delivered to
Google. However, the union claims that given the sale agreement requires links across data
sets to be preserved, it causes concern that individual information could be reconstructed.
As for Google itself, they plan to use the data for product development and training of its AI
models.
The most obvious fear is that, once more, a tech giant will be using third-party data to
develop and train its Large Language Models to ultimately automate people’s jobs away.
But in addition to that, and the risk of compromising employees’ privacy, the sale could
underpin something more nefarious.
The move potentially opens the likelihood of other companies who are facing financial
difficulties—or simply seeking new forms of revenue—could decide to sell to tech firms the
internal and employee information they don’t consider critical to their business.
But non-profit digital rights organizations like the Electronic Frontier Foundation (EFF) argue
that neither solvent nor bankrupt firms should be allowed to market their data unless
individual employees who generated that data agree.
However, another reason many tech ethicists and business observers are wary of the Spirit
deal is the concern regarding the implications of AI tech being trained with vast and
detailed data covering virtually every aspect of a third-party company’s operations.
With that, AI can be increasingly trained to run business units, not just a job.
That, in turn, would theoretically allow Google and other AI developers to teach bots not
only how to replicate individual employees’ output, but would likely allow them to
understand and reproduce how their teams, departments and even entire companies
function while running that business.
This mirrors Meta’s move last April to install software that would track all employees’
actions—every mouse click, key stroke, screen shots–to collect real world data to train
autonomous AI agents. Heavy pushbacks from staff and internal protests saw Meta scale
back the operation, ultimately suspending it after a security error left sensitive data
available for anyone in the company to view.
Over the last 18 months, one of the biggest trends in AI is the rise of “reinforcement
learning (RL) environments”—a method of AI training that has led to rapid improvements in
models that can work independently for long periods of time. These so-called RL
environments are essentially simulations of commonly used software, where AI agents can
take strings of action independently. If the action leads to the desired outcome, the AI is
rewarded and will be more likely to take similar action in the future.
Still, it’s worth noting that reinforcement learning environments are only as good as the
data that populates them. And while Spirit was a failed company, the data still encodes
useful things about how humans carry out their jobs, like how workers collaborate toward
shared goals, or how the norms of the aviation industry operate.
But unlike coding, which can quickly be determined whether it works or not, office tasks
are much more nuanced, and so it is unclear whether populating RL environments with real
world data like Spirit’s will allow AI companies to move into white-collar fields as quickly as
it has through the software industry.
Be the first to comment