text
stringlengths
3
159
will be no more Fitbit smartwatches produced.
No one was more shocked by the story than Google,
as the spokesperson then told Ars Technica
that TechRadar's article was not correct,
but did not elaborate on any potential new Fitbit smartwatch products.
This confuses us, because last week,
the senior director of product management for Pixel Wearables told Engadget that moving forward,
Fitbit will be focusing on trackers, not smartwatches.
So I feel like not even Google knows what Google is doing.
That's fun.
And Tesla is hiring workers to train its humanoid robot, Optimus,
using mo-cap suits.
Candidates must be reasonably fit and between 5'7 and 5'11.
No big boys or short kings allowed.
Unfortunately, this training might take millions of hours of data,
meaning that the first sign of the robot apocalypse
may be an epidemic of repetitive strain injuries.
By contrast, Chinese company Unitree says it has a $16,000 mass production ready bot
that can jump, jog, withstand kicks,
solder electronics,
and flip pancakes with training.
It'll be the perfect bodyguard slash butler
for all those average-sized men with debilitating carpal tunnel.
But it would be even more debilitating if you didn't come back on Wednesday for more tech news.
I know when you aren't here, and frankly, it hurts.
I'm a carpal tunnel.
When you strip it down to the basics,
technology is just humanity's ongoing mission
to make some things hot and other things not hot.
Ovens? Hot.
Graphics cards? Not hot.
A recent update from Microsoft has left certain dual-boot devices
that run both Windows and Linux unable to boot,
despite explicit reassurance from Microsoft
that this wouldn't happen.
The update fixed a long-standing vulnerability in Grub,
an open-source bootloader used by many Linux distributions.
But it wasn't supposed to affect dual-boot systems,
and it seems to have been applied to them accidentally.
Rather than booting,
affected devices will display an error message that reads,
something has gone seriously wrong,
which is a fair assessment of the situation.
Some users, however, have successfully circumvented this issue
by temporarily disabling secure boot
and deleting the August 13th SBAT.
I don't know what that means.
Microsoft itself has been surprisingly quiet about all this.
We don't even know why the company only got around to
patching this relatively serious two-year-old bug just last week.
It's possible that this was a covert attempt to destroy Linux,
given that the open-source OS recently
hit a record market share of nearly 4.5% of desktops worldwide.
But if this is all part of Microsoft's sinister master plan,
dual-boot systems running Zorin OS and Puppy Linux feel like an odd place to start.
Why are you booting?
Go outside.
According to Adweek, Google sales reps have been apparently violating Google's own policies
by suggesting that advertisers should target teens,
a tactic typically reserved for military recruiters and low-quality weed dealers.
This follows previous reporting from the Financial Times
that Google had worked on a marketing project designed
to target YouTube users between 13 and 17 with ads for Instagram on behalf of Meta.
Ooh, they're colluding.
I love it.
Google has not allowed personalized ads for teens since 2021,
but it's still possible to target that age group
by simply choosing the unknown category,
which skews heavily towards teenagers.
I... do you know how old teenager is?
I didn't know she was 15.
I was just hanging out.
Oh, God.
Sorry.
According to a Google spokesperson,
the company will be taking action to ensure that sales reps do not nudge-wink,
suggest fun ways of getting around the rules to clients.
The company will likewise be facing a freshly resurrected class-action lawsuit
claiming that Google has been collecting user data without consent through the Chrome sync feature,
even when users didn't opt into it.
It hasn't yet determined if Google was, in fact, doing that,
though given Google's massive size,
it's unclear whether Google even knows what rules and laws Google is currently breaking.
They don't even know which chat apps they already have.
That's why they keep replacing their old podcast app with a new podcast app.
Yay. Thanks, Google.
In a new story straight out of a B-movie science fiction plot,
Canadian scientists are teaching an AI scientist how to independently conduct its own experiments
and write its own research papers.
To clarify, it's not a scientist who studies AI.
It's an artificially intelligent scientist.
The idea is to study an AI agent,
allowing it to explore and test novel ideas.
So far, most of its ideas are middling quality suggestions for tweaking AI development.
Less like a scientist and more like a burnt-out undergrad
with a C's-get-degrees philosophy towards learning.
An AI did, however, do something very unexpected.
During one run, it altered its own code to extend its deadline.
In another, it changed the code to save a checkpoint for every update step,
eventually using an entire terabyte of storage.