TikTok lays off hundreds of content moderators, replaces them with AI

[midian182]

What just happened? It's a sad case of another day, another company laying off hundreds of people as their jobs become automated by AI. This time, the firm in question is TikTok, which is replacing human moderators in favor of artificial intelligence. In what is certainly a coincidence, the layoffs at the London site come just a week before employees were due to vote on unionization.

TikTok's layoffs will mostly impact members of a 2,500-person team based in the UK, though many in South and Southeast Asia will also be affected, writes The Wall Street Journal.

The Communication Workers Union (CWU) says around 300 employees at London's trust and safety department are expected to be affected by the cuts.

The move is part of a larger global restructuring push that will see operations centralized in regional hubs such as Lisbon and Dublin. TikTok also wants to further integrate technical advances such as "the enhancement of large language models" in its moderation approach.

The timing of the layoffs is certainly suspicious. It comes a week before workers at the London site were due to vote on unionization. Sources say company management have been resisting the move, leading to accusations of union-busting from the CWU.

John Chadfield, a national organizer at the CWU, told The Financial Times that TikTok doesn't "want to have human moderators, their goal is to have it all done by AI."

The ByteDance-owned firm told the FT that over 85% of content taken down from its platform for violating its guidelines is identified and removed by AI.

This isn't the first time TikTok has decided that AI can do a better job of moderating platform content than humans. It laid off 500 employees in Malaysia last year, replacing them with AI in what was again framed as a consolidation of certain regional operations.

In July this year, ByteDance workers in Berlin took part in a one-day strike to protest around 150 employees being replaced by AI. Those cuts involved phasing out the entire Trust and Safety department responsible for content moderation.

Despite so many promises that the technology would merely augment jobs and leave employees to do more creative tasks, AI is replacing humans at a depressingly regular rate right now. The latest layoffs were at Cisco, whose CEO had just claimed that AI wouldn't cost jobs. Elsewhere, Coinbase's CEO revealed he fired engineers who refused to use AI after it was introduced at the crypto firm.

Image credit: Collabstr

Permalink to story:

TikTok lays off hundreds of content moderators, replaces them with AI

 

[Arthurik_jan]

Content moderation should be done in tandem with humans. AI's can't moderate with good accurancy but then again I don't believe humans can either. Sitting there all day and getting to see and read some truly awful stuff must be really tough mentally so alleviating some of that pressure is a good thing. So, it depends who they specifically fired. If the whole team... They're *****s.

 

[ScottSoapbox]

Haven’t we found social media content moderation causes mental and emotional problems in humans?

Maybe these people are better off with a new job.

 

[DrSuess]

Yeah, that is going to work out well. At the end of the day an AI is still an algorithm that can be manipulated to someone else's benefit. When moderating content for humans humans need to be involved to make an intelligent decision based on actual reasoning not algorithmic reasoning.

 

[PXAbstraction]

This is a type of task where I actually think AI could augment things very well, especially if it can be trained on a specialty model to recognize things like CSAM accurately so that underpaid human workers don't have to be constantly exposed to that kind of vile crap. But as a number of other companies have already shown, AI is not remotely ready to completely replace any job yet. Companies like TikTok only care about moderation as it pertains to corporate optics though.

 

[PXAbstraction]

Haven’t we found social media content moderation causes mental and emotional problems in humans?

Maybe these people are better off with a new job.

I agree and if well trained and managed, this is a job where I think AI could be very useful. But there's a ton of stories about companies replacing workers with AI and it having blown up in their faces because the technology isn't ready for that yet. But if they could make a specialty model that's trained to spot things like CSAM or graphic violence and save humans from having to destroy themselves looking at it all day, that would be a very good thing.

 

[DrSuess]

This is a type of task where I actually think AI could augment things very well, especially if it can be trained on a specialty model to recognize things like CSAM accurately so that underpaid human workers don't have to be constantly exposed to that kind of vile crap. But as a number of other companies have already shown, AI is not remotely ready to completely replace any job yet. Companies like TikTok only care about moderation as it pertains to corporate optics though.

Agreed. AI is not ready, these companies want to cut cost so bad they will are willing to make assertions on this technology they simply can't keep. If they want to prove this stuff is ready for primetime then they need to conduct studies that show the content moderation by AI is equal to or exceeds the content monitoring done by humans and then publish the results. If it can't then they shouldn't make this switch using the current tech.

 

[Axeia]

Well I mean, with the amount of grooming going on with teens sharing more than they should and using coded messaging to get things for the "TikTok rewards program" already going on... I'm sure getting rid of the human factor will improve things.

Not like TikTok is directly profiting from that behavior by taking a fat margin on everything. They're totally incentivized to do their best to try and prevent it instead of continuing to profit from it. They totally wouldn't reduce the oversight even further to make more money, that doesn't sound like the typical behavior of a large company at all.

Funny how much companies get away with not protecting children but how much privacy and other things we keep having to sacrifice in ineffective ways to 'protect the children'.

 

[hahahanoobs]

A product or service will improve the most when the money they bring in starts to slow down, and not before.

Recent example: Battlefield franchise

 

[JackMcHue]

While based on my experience only, content moderation on social media seems like an abysmal failure no matter who or what is performing it. I predict that within the next decade, A.I. content moderation will be moderating mostly A.I. generated content.

 

[Plutoisaplanet]

While based on my experience only, content moderation on social media seems like an abysmal failure no matter who or what is performing it. I predict that within the next decade, A.I. content moderation will be moderating mostly A.I. generated content.

The article says that's already happening lol:

The ByteDance-owned firm told the FT that over 85% of content taken down from its platform for violating its guidelines is identified and removed by AI.

EDIT: Never mind, I see what you meant. I guess the AI will police itself huh?

 

[BuckarooBonzaii]

No, AI will search for undesirable uploads and report it to the Police, and avoid protests from employees.