Big Dog takes to the woods

[Red]

Just burns the tops off.  Many weeds are fully capable of continuing to grow unless you kill the root mass.

[Joe Frickin' Friday]

7 hours ago, Red said:

Just burns the tops off.  Many weeds are fully capable of continuing to grow unless you kill the root mass.

 

I get that some weeds can't be permanently killed unless you get most/all of the roots. But it's hard for me to imagine the roots making much progress when the greens get completely incinerated every few weeks.   I guess the farmers will know for sure whether this thing has any value or not.

 

A better video of the weed-roasting action:

 

 

 

[szurszewski]

Slap that in a drone with facial recognition…what could go wrong?

[Rougarou]

U. S. Robotics are here, Skynet ain't that far behind!!!  But I'd take an Ex Machina model anyday

 

 

[szurszewski]

nothing terrifying about that at all....

[Joe Frickin' Friday]

On 12/2/2021 at 6:12 PM, Rougarou said:

U. S. Robotics are here, Skynet ain't that far behind!!!  But I'd take an Ex Machina model anyday

 

 

 

It moves like a real human being mired in a tub of molasses.

[BamaJohn]

Not sure where I saw this, but it absolutely amazed me...See what you think:

[wbw6cos]

Taking a chance on not having a computer glitch.  Wow.

[Hosstage]

No way that's real. If so, scary as hell. The handlers are more brave than any man I know.

[TEWKS]

Was a neat show but trickery was used.

 

[Hosstage]

That makes more sense.

[BamaJohn]

Yes, it does!  

[Joe Frickin' Friday]

The robot claims it was an accident...

 

Chess robot breaks seven-year-old's finger during Moscow Open

 

 

 

[Hosstage]

 

[Joe Frickin' Friday]

I recently wrote about tinkering around with an AI image generator.  I was using MidJourney, but that's not the only one out there.  There is also Dall-E (originally named for the surrealist artist, but recently renamed as Craiyon), more recently there's Dall-E 2, NightCafe, and others.  I've only tried MidJourney a few times using the simplest of inputs, but signing up let me also see a livestream of what other users were cranking out.  If you know what you're doing, you can offer up some pretty complex input and get some really amazing output.  People also have a wicked sense of humor, as the Dall-E subreddit shows.  I scrolled through it, and laughed at these two:

 

 

 

There's also a MidJourney subreddit, with entries like Van Gogh Travels to Kyoto, and Gundalf the Pump Action Wizard.

 

So why am I bringing this up in a thread about robots?  Glad you asked:

An AI-Generated Artwork Won First Place at a State Fair Fine Arts Competition, and Artists Are Pissed

 

As the title suggests, someone used MidJourney to create an image and won first place in the digital category at an art contest.  Gotta admit, it's a pretty impressive image:

 

 

Digital art is nothing new, but until recently, it was made using more basic editing tools.   But artists are aghast at this guy winning an art contest using an AI image generator, bemoaning the death of their profession:

 

Quote

“We’re watching the death of artistry unfold before our eyes,” a Twitter user going by OmniMorpho said in a reply that gained over 2,000 likes. “If creative jobs aren’t safe from machines, then even high-skilled jobs are in danger of becoming obsolete. What will we have then?”

 

I imagine most of these artists don't complain about the fact that their carrots and potatoes were harvested, cleaned and packaged by machines, or that their brushes, paints, colored pencils, computers and monitors were all produced in highly automated factories.  But now they're concerned because AI image generators greatly lower the bar for creating pleasing, high-quality imagery, potentially putting skilled artists out of work.  And they're right.  As the article points out, AI-generated imagery has already been used in circumstances where an artist or photographer might previously have been paid to provide an image.  And as AI image generators get better, more and more artists - digital and otherwise - will find themselves with less and less paid work, just like the folks who used to pick carrots and potatoes.    

 

[chrisolson]

4 hours ago, Joe Frickin' Friday said:

as AI image generators get better, more and more artists - digital and otherwise - will find themselves with less and less paid work

 

Will there be an interesting turn around .... where 'human generated' art becomes the novelty ... and / or potentially see that human skill dying out ?

[Hosstage]

3 hours ago, chrisolson said:

 

Will there be an interesting turn around .... where 'human generated' art becomes the novelty ... and / or potentially see that human skill dying out ?

I think unfortunately to a big degree, yes.

 

[Rougarou]

FN Meka,......music, lyrics were supposedly AI created.  Only the voice was human, but even that can be generated.

 

How soon to all entertainment is generated by AI.

 

 

[taylor1]

Augmented reality is also a very cool tech. My son has written software for Activision and now for Apple. A year ago he sent me a video text he made. This is an app they are working on. It was his wife, son, and dog playing in their living room. All of a sudden, a small dinosaur runs into the room, grabs their dog and runs out. It was so real looking it kind of freaked me out    LOL

[Hosstage]

So video will no longer be a mostly trusted form of evidence in court. Great.

There are also very few "artists" on the current pop scene that can actually sing, almost all is now heavily auto-tuned, annoying as hell.

[Rougarou]

41 minutes ago, Hosstage said:

So video will no longer be a mostly trusted form of evidence in court. Great.

There are also very few "artists" on the current pop scene that can actually sing, almost all is now heavily auto-tuned, annoying as hell.

 

Nope, video cannot be trusted

 

 

 

 

And if you, uh hum, peruse the adult educational videos of certain sites, you can easily be confused by the deep fakes.  I'd post video links, but that may get me a nasty time out.

 

[Hosstage]

1 hour ago, Rougarou said:

 

Nope, video cannot be trusted

 

 

 

 

And if you, uh hum, peruse the adult educational videos of certain sites, you can easily be confused by the deep fakes.  I'd post video links, but that may get me a nasty time out.

 

I did happen to catch that episode of AGT. Creepy.

[Joe Frickin' Friday]

EDIT: I wrote this earlier this afternoon, but got distracted.  I see I've been scooped on the deepfake issue, but I'll go ahead and get this out there anyway...

 

5 hours ago, Hosstage said:

So video will no longer be a mostly trusted form of evidence in court. Great.

 

Deepfake tech, able to create convincing video of real people, has been around for a few years now.  The big thing holding it back has been the need for a lot of processing power to get good quality results in a timely fashion.  But algorithms and processing power have been improving, and this video shows you can now get pretty good results in real time:

 

 

You can spot the defects in those results, but again, those were generated in real time, i.e. no delay between the actions of the real person and the responses of the deepfaked video image.  If you give the software some time work on it, it's not hard to imagine getting results that appear real to the naked eye.  There are (or at least, there were) video analysis techniques capable of spotting deepfakes, but as the technology improves, there will be fewer (and smaller) artifacts for detection schemes to detect.  And then, yes, we will have arrived at a point where video evidence will no longer be trusted.

[Red]

It won't be long before major internet protection companies will have anti deep fake software for sale.

[Joe Frickin' Friday]

A little further down the path of "robots that can be useful or unexpectedly dangerous, we get this:

 

Scientists Create AI-Powered Laser Turret That Kills Cockroaches

 

Quote

In the study—which was conducted last year but published in Oriental Insects last week—Rakhmatulin and his co-authors used a laser insect control device automated with machine vision to perform a series of experiments on domiciliary cockroaches. They were able to not only detect cockroaches at high accuracy but also neutralize and deter individual insects at a distance up to 1.2 meters. 

.

.

.

According to the paper, Rakhmatulin tried this configuration at different power levels for the laser. At a lower power level, he found that he could influence the behavior of roaches by simply triggering their flight response with a laser; this way, they could potentially be trained to not shelter in a particular dark area. At a higher power level, the cockroaches were effectively "neutralised," in the paper's language—in other words, killed. 

.

.

.

It’s also sadly not quite ready for household use, at least not yet. “It’s not recommended because it’s a little dangerous,” Rakhmatulin said. “Lasers can damage not only cockroaches but your eyes.” 

 

 

 

 

[mnTwin]

3 hours ago, Joe Frickin' Friday said:

A little further down the path of "robots that can be useful or unexpectedly dangerous, we get this:

 

Scientists Create AI-Powered Laser Turret That Kills Cockroaches

 

 

 

 

 

“It’s not recommended because it’s a little dangerous,” Rakhmatulin said. “Lasers can damage not only cockroaches but your eyes.”

 

Understatement of the day. (Eyes, skin, other non-beam hazards such as breathing the vaporized remains of the "neutralized" roaches!) That reminds me, I have laser safety training coming up again at work soon. Having said that, it would be really cool. (But I can think of way too many ways it could fail badly.)

[John Ranalletta]

Knock, knock.  Is anyone home?
 

 

[Joe Frickin' Friday]

5 hours ago, John Ranalletta said:

Knock, knock.  Is anyone home?
 

 

 

Well, the military already has airborne drones that can shoot to kill, so a walking drone isn't really hugely different.  Don't even need AI to make action decisions - just good fundamental control algorithms to keep it walking upright over varied terrain while a remote operator tells it where to move and what to do.  

[Joe Frickin' Friday]

You think you're safe up on that scaffolding?  Think again.  Atlas will find a way up, and even if it can't get all the way to you, it can throw stuff to try to knock you off:

 

 

 

 

Background on what went into the making of that video, and how Atlas perceives its world and how it plans and executes its actions:

 

 

[Joe Frickin' Friday]

While Big Dog and Atlas are coming for the blue-collar jobs (and possibly also for your lives), AI is also coming for the white-collar jobs.  Code monkeys beware: ChatGPT is in the news lately for its ability to write coherent articles/essays, but that's not the only thing it can do: it can also write functional software programs based on some simple plain-text inputs from a user.  

 

 

 

ChatGPT Passes Google Coding Interview for Level 3 Engineer With $183K Salary

Quote

Google fed coding interview questions to ChatGPT and, based off the AI's answers, determined it would be hired for a level three engineering position, according to an internal document.

 

As reported by CNBC, the experiment was done as part of Google's recent testing of multiple AI chatbots, which it's considering adding to the site. ChatGPT's ability to surface a concise, high-fidelity answer to a question could save users time typically spent surfing links on Google to find the same information.

 

“Amazingly, ChatGPT gets hired at L3 when interviewed for a coding position,” says the document. And while level three is considered an entry-level position on the engineering team at Google, average total compensation for the job is about $183,000.

 

 

 

 

[Rougarou]

[szurszewski]

3 hours ago, Joe Frickin' Friday said:

While Big Dog and Atlas are coming for the blue-collar jobs (and possibly also for your lives), AI is also coming for the white-collar jobs.  Code monkeys beware: ChatGPT is in the news lately for its ability to write coherent articles/essays, but that's not the only thing it can do: it can also write functional software programs based on some simple plain-text inputs from a user.  

 

 

 

ChatGPT Passes Google Coding Interview for Level 3 Engineer With $183K Salary

 

 

 

 

I watched that yesterday (or the day before? On the couch last since Friday with dysfunctional knee - hard to keep track of which day was which) - I found it very interesting. 

[chrisolson]

The disciplines that can/will be deeply affected by this technology are unknown .... but that's his point.  There have already been interesting applications in medicine such as creating new enzymes ... what's next ?

 

I think he's right that we are early in the curve and have no idea of where its going to take us ... particularly with the massive investment of corporations like Microsoft.  Some of it may have definite benefits to humanity and others maybe not so much.  Likely It will be difficult in the future for the average person to distinguish between machine generated content and human ... regardless of what that content is.

[Joe Frickin' Friday]

Early attempts at AI image generation were pretty bad, but the tech improved rapidly and we're starting to see some really impressive results lately.

 

Let's hope we see the same rapid improvement for AI video generation, because man, these early results are freaking nightmare fuel:

 

 

 

[doc47]

On 10/23/2013 at 11:48 AM, Matts_12GS said:

 

It's all well and good till they run out of other people's money

A song by Leon Rosselson:

Come to the market and sample an orgy
Come to the fair for a thrupenny thrill
Sell him a tomb or a tumbler of strychnine
If you don't there's plenty of others that will
If you don't there's others that will

But what shall we do with the ugly ones
The ones who have nothing to sell
The failures the fumbling muddly ones
Who never do anything well
Who never remember their name or number
And lose their place in the queue
And what can you do for the ugly ones
When they can't do a thing for you

Roll up roll up to the mechanised peepshow
Bow down bow down in the temple of lust
What am I bid for the lips of a Bardot
And Anita's marvellous marvellous bust
And Anita's marvellous bust

But what shall we do with the ugly ones
Who just haven't got what it takes
Whose breasts won't boost the marketing charms
Or win the nubility stakes
Their legs won't sustain the ad man's campaign
Or front the glamour parade
And no one invests in the ugly ones
The ones who won't make the grade

Dress yourself smart for the paysetter's party
Dance to the swing of the trendsetter's call
The prince is cool in an Aston Martin
Eying Cinderella she's the belle of the ball
Cinderella's the bell of the ball

But what shall we do with the ugly ones
The crippled the sick and the old
Who haven't got anything left to do
But shroud themselves off from the cold
Give them a penny - they haven't got any
It's time for the charity game
But we can't change the rules for the ugly ones
And nobody here's to blame

The moth-eared midget is starting to fidget
Soon it will be his turn to go
The flesh and the fur are starting to stir
Hurry up dear or we'll miss the show
Be quick or we'll miss the show

But what shall we do with the ugly ones
The freaks with nothing to sell
The stupefied stunted shell-shocked ones
In their halitosic hell
They can't stand the pace of the status race
Or cash in on the rush to rebel
And there's nothing to do for the ugly ones
The ones with nothing to sell
 

[Joe Frickin' Friday]

AI image generators are getting pretty good at generating images with photorealism, but they still sometimes misunderstand what people are asking for:

 

[Joe Frickin' Friday]

On 4/1/2023 at 7:02 PM, Joe Frickin' Friday said:

Early attempts at AI image generation were pretty bad, but the tech improved rapidly and we're starting to see some really impressive results lately.

 

Let's hope we see the same rapid improvement for AI video generation, because man, these early results are freaking nightmare fuel:

 

 

 

 

Just about a year ago, Will Smith was eating spaghetti in your nightmares.  AI video generation isn't exactly indistinguishable from reality just yet, but it's gotten surprisingly good in a very short time.  Here's an AI-generated movie trailer for The Matrix, except it looks like a vintage sci-fi/spy thriller.  Even the narration script was produced by AI.  There's a really interesting recursion in using AI to generate a realistic trailer for a movie about an AI that dupes people into accepting a fake version of reality.

 

 

 

[Skywagon]

Salmon in river… just add some wasabi and instant no fuss 🍱 sushi 

[BrianT]

I actually like that retro 50's or 60's take on the Matrix.  Make it less action, more horror or scare and it would fit right in to the Science Fiction movies that came out during that time period.  Looks better than any of the Matrix sequels they put out.

[Joe Frickin' Friday]

This week, Boston Dynamics released a video showing the latest version of Atlas, with joint mobility that will make it the envy of contortionists everywhere:

 

 

 

In case it's not obvious, the new Atlas is all electric, whereas the old one was hydraulic.  BD has also posted a retrospective, showcasing the old HD Atlas robot's greatest hits and misses:

 

[Rougarou]

The way the arms go to the mid section, is like the 'bot hurt its bolt and nuts in the fall

 

 

 

 

 

In some of the other videos, it appears to do the same thing,.....thought it kinda funny that it must be a male 'bot.

[John Ranalletta]

On 12/2/2021 at 6:12 PM, Rougarou said:

U. S. Robotics are here, Skynet ain't that far behind!!!  But I'd take an Ex Machina model anyday

 

 

 

Put one of these on a MotoGP bike with all the telemetry (tire temps, speed, circuit map, lidar, etc.) going directly into the robot's cpu coupled with no fear of death.

[Joe Frickin' Friday]

1 hour ago, Rougarou said:

The way the arms go to the mid section, is like the 'bot hurt its bolt and nuts in the fall

 

 

A number of people in the video comments noticed the same thing.  I wonder if it was a deliberately programmed action to protect the arms from damage once Atlas determined that a fall was unavoidable.  There are a couple of clips (see at 1:01, 1:04, 1:27 ) where Atlas blows a hydraulic leak in a leg joint, so it seems like they were operating it right at the limit of its capability to do those acrobatic stunts, and maybe they knew the arms weren't strong enough to be useful for bracing against a fall - just let the armored main body take the impact instead.  

[Joe Frickin' Friday]

someone proposed giving the new electric Atlas a fur suit:

 

 

Not sure if it's better or creepier.

[Joe Frickin' Friday]

The latest step toward Skynet:

 

 

I wonder when we'll hand over fire control authority.  

 

[Hosstage]

5 hours ago, Joe Frickin' Friday said:

I wonder when we'll hand over fire control authority.

Skynet don't need no stinking authority.

[Brucifer1150]

On 4/20/2024 at 9:31 AM, Joe Frickin' Friday said:

someone proposed giving the new electric Atlas a fur suit:

 

 

Not sure if it's better or creepier.

 

Already done:
 

[Brucifer1150]

On 11/9/2021 at 10:55 AM, szurszewski said:

Slap that in a drone with facial recognition…what could go wrong?

 

 

[BrianT]

For $9,280, you too can have your very own flamethrower robotic dog.  That's almost as good a sharks with frickin' lazer beams.  Fast forward to 52 seconds into the video if you're in a rush to get to the fiery parts.

 

 

[Hosstage]

I want one!

Hey you kids, get off my lawn!

Whattya gonna do old man, sic your dog on us?

You do not want me to sic my dog on you...