Artificial intelligence still has some way to go

wait, so he would just be standing on the hotel bed while the sex workers pretend to get peed on? i can see why he would want to that on tape

ok I laughed at "Orlando Brooms"
https://64.media.tumblr.com/21a25798ac613df3f73050def1d3088b/3812fd1d9f78030b-3f/s500x750/9edc035893332b6682f312b8c65592c8224311d0.png

from https://aiweirdness.com/post/633411394686042112/splorts-teams

https://image.ganref.jp/photos/members/wutaibo/2ba3758a04920247790b47410f36c916_3.jpg

The club announced a few weeks ago it was moving from using human camera operators to cameras controlled by AI. The club proudly announced at the time the new "Pixellot system uses cameras with in-built, AI, ball-tracking technology" and would be used to capture HD footage of all home matches at Caledonian Stadium, which would be broadcast directly to season-ticket holders' homes.

Cut to last Saturday, when the robot cameras were given a new challenge that hadn't been foreseen: A linesman with a bald head.

Inverness Caledonian Thistle don’t employ a cameraman as their camera is programmed to follow the ball throughout the match. The commentator had to apologise today as the camera kept on mistaking the ball for the linesman’s head... pic.twitter.com/LeKsc2bEj7
— Tom Cox (@seagull81) October 24, 2020

The AI camera appeared to mistake the man's bald head for the ball for a lot of the match, repeatedly swinging back to follow the linesman instead of the actual game. Many viewers complained they missed their team scoring a goal because the camera "kept thinking the Lino bald head was the ball," and some even suggested the club would have to provide the linesman with a toupe or hat.

https://www.iflscience.com/technology/ai-camera-ruins-soccar-game-for-fans-after-mistaking-referees-bald-head-for-ball/

if anything has ever really needed to be polled, it is Blaseball: The Simulation

https://www.iflscience.com/technology/ai-camera-ruins-soccar-game-for-fans-after-mistaking-referees-bald-head-for-ball/

Brilliant.

and the award for creepiest work email yet received goes to...

"This Alteryx workflow has become self-aware and will now send you work automatically..."

I tweeted (alright, I moaned) about how Amazon ai sucks because they sent me an email advertising something I literally just bought. (Important note: I didn't @ them - just mentioned "Amazon"), and their twitter bot instantly responded with a helpful tip on how to deactivate the market emails.

so i guess it sucks and doesn't suck at the same time.

and yeah, my tweets are a barrel of laughs.

I am driven mad by Amazon emailing me ads for products I have bought from them and they cancelled the order because they couldn't fulfil it. Fucking dumbfucks.

My PS4 keeps recommending shows I'm already watching on my PS4. HOW DOES IT KNOW???!!

I'm pretty sure those promo thumbnails hook up to the various services' algorithms if you're logged in on your PS4

i mean, if you're logged into those services

It asked for permission to monitor my video app activity in order to make recommendations then it recommends things after I've already discovered them.

They probably bought a used spotify algorithm off ebay.

lol this is pretty weak:
https://www.youtube.com/watch?v=RKiGfGQxqaQ

they'll get to Max Headroom in around 50 years at this rate

for companies that want something 800% more disturbing than a real human, while only being 500% less helpful

the trend to name AIs after "normal" names (Douglas, Watson, Erica, etc) is weird. it feels like a new dot com era domain buy-up spree. similar to how right now we're all like "why didn't *I* register awesomecars.com in 1998, i would be rich now!", there may come a time when advanced AI makers wish that they could name their creations Greg

Like, I would understand why DD would want realistic models of dead actors, etc, but why on earth would I want to make one of myself?

tbf he does look and move a lot like aged paul ryan. so, almost kinda lifelike sorta.

"As lifelike as Paul Ryan" you mean.

trying to avoid oxymorons when possible

The glaring problem with AI 'Douglas' is that, even though the computer apparently has been programmed to be able to reproduce the face, posture, or gestures of its model human in 'every conceivable position', the computer has zero idea which of the millions of potential micro transitions to its virtual face and body that would be most appropriate to smoothly mimic a real human responding to unscripted input. It just jumps discontinuously from one unconvincing posture and expression to another. iow, it has excellent graphics and crap AI.

https://www.youtube.com/watch?v=P4IViUHEPNs

loooooool

that one hurt

This is impressive: https://www.independent.co.uk/life-style/gadgets-and-tech/protein-folding-ai-deepmind-google-cancer-covid-b1764008.html

https://www.youtube.com/watch?v=8HZ4DnVfWYQ

closed captions recommended there

fractal cucumber!

https://www.youtube.com/watch?v=gg7WjuFs8F4

The Biden-Harris administration is committed to rebuilding an economy that welcomes everyone as full participants.

— The Democrats (@TheDemocrats) December 1, 2020

It's great that journals are getting authors to do plain language summaries along with the more technical abstracts.

This one - from an article titled 'Potential and limitations of machine learning for modelling warm-rain cloud microphysical processes' - is plainer than most. 🙂 https://t.co/UyKoDvQPtT pic.twitter.com/W3a1OCrBxs

— Tim Baxter #athome (@timinmitcham) December 5, 2020

why haven't scientists always been doing that, omg

terrific

At first I thought an AI wrote that and I was like “I guess they don’t have that far left to go!”

This seems to be the thread previous Google web audio experiments were on, and this involves machine learning though I don't know exactly how, so, bloody hell, this is impressive, and also cute:

https://artsandculture.google.com/experiment/blob-opera/AAHWrq360NcGbw

Control your own blob choir! 4-part harmony! Sings Christmas carols (click the 🎄)!

Introducing Blob Opera, an experiment I created for Google Arts & Culture: https://t.co/dfLnCXSR0R pic.twitter.com/UYbZH8jXE2

— David Li (@daviddotli) December 15, 2020

That's pretty neat!

yeah, i love it

Ohh, apparently it also takes midi input. Which is also an excuse to post this Satie gymblobédie.

Discovered that @daviddotli's awesome Blob Opera project works with my MIDI keyboard! 🎹 Enjoy these blobs singing Satie. 🎵 @googlearts pic.twitter.com/6ZKHaK2o4X

— Alexander Chen (@alexanderchen) December 17, 2020

A big reason why these blobs are so wonderful is they are not the source of the intelligence which is making the music. That's being supplied by real musicians.

I kinda want to use that to make some fake vocals for Bicep-sounding tracks.

neat!
https://openai.com/blog/dall-e/

pika pika pic.twitter.com/ErIF703OBb

— Max Woolf (@minimaxir) January 5, 2021

gotta AI fast pic.twitter.com/myb8oPzVuZ

— Max Woolf (@minimaxir) January 5, 2021

OpenAI’s new DALL-E is GPT-3 but trained on language + images. It can generate pictures from text that blend concepts in interesting ways https://t.co/CtFwPcJ5L6 pic.twitter.com/wxDzIExqMX

— Mark O. Riedl (@mark_riedl) January 5, 2021

a human worker who is apathetic and slow, replaced with a machine that is never tired

a honking eggplant begging for licks

very much want to test this out

saaaaameeeeee let me in let me in let me in

your mom

https://venturebeat.com/2021/01/11/outlandish-stanford-facial-recognition-study-claims-there-are-links-between-facial-features-and-political-orientation/

Kosinski and coauthors, preemptively responding to criticism, take pains to distance their research from phrenology and physiognomy. But they don’t dismiss them altogether. “Physiognomy was based on unscientific studies, superstition, anecdotal evidence, and racist pseudo-theories. The fact that its claims were unsupported, however, does not automatically mean that they are all wrong,” they wrote in notes published alongside the paper. “Some of physiognomists’ claims may have been correct, perhaps by a mere accident.”

You may recall this is the same discredited charlatan Stanford prof who claimed AI could detect sexual orientation.

very much want to test this out

In my experience it is simultaneously true that

a) systems like this show very impressive performance;
b) what you see in the demo is the cream of the crop and if you mess around with it yourself you'll quickly come to see what kinds of tasks it's good at doing and a much larger universe of superficially similar tasks it's not good at all at doing