Post

Would be nice if people stopped restating the false claim that Waymos have live humans driving them. That was based on a misleading Futurism headline that was debunked in its own article. There is way too much hype about AI tech but it does no good to misrepresent the state of things the other way.

This post is unavailable.

30 reposts249 likes

IDK what that now delete post said, but Waymo's Passenger Safety Plan says they have remote operators who can "reposition" their robots at low speed and their letter to Sen Markey appears to confirm that. There are multiple videos that sure look like, eg: bsky.app/profile/anic...

John Berry@aniccia.bsky.social

"I'm putting it into manual controls right now" Waymo telop, who then remotely drove a robot w passenger in Los Angeles. Apparently, the telop "assisted" the robot to move a little, then manually tookover for a tricky left maneuver. Video in 2 parts bc of bsky 3 min limit. OP: tiktok.gstacks_sama

Yes. They have remote operators who step in the rare cases of something going wrong, who then can help get the car to safety. But there was a story making the rounds a week or two ago implying that ALL waymos are driven ALL THE TIME by remote operators.

Sure, exaggerations both ways. Waymo said their robots are engaged w telops ~5% of the time, which is not "rare" for a lifesafety system. They do it routinely for many ordinary conditions, eg traffic light outages which swamped their telops affecting SF on 12/20/2026. bsky.app/profile/anic...

John Berry@aniccia.bsky.social

If ~5% of the time on ave Waymo robots are stuck/MRC/failed needing remote assistance, then they are ~1.5 nines reliable (~95%). Typically, lifesafety systems are 5-6 nines reliable. Human drivers average in this range, inc DUI/teen/etc and ODD/conditions way too tough for Waymo's robots + telops.

Sadly, lots of people only believe things that confirm their priors. It's why Fox News exists, it's how Trump got elected telling a relentless stream of lies but it's certainly not confined to people on the right side of the spectrum...

Worth noting: the workers in India reported to be tracking Amazon Go purchases were mostly data labelers — annotating visual scenes to train the computer vision models behind cashierless checkout. It's a critical (and often invisible) part of how visual AI actually gets built.

sometimes futurism seems like an okay source for tech news i don't see elsewhere, as they seem obsessed with finding every little thing musk does wrong, for example. but they also promote the space pseudoscience of avi loeb so they can't really be trusted.