AI beer commercial


mikbone
 Share

Recommended Posts

Also for the record, a few weeks ago I gave probably my best talk in Sacrament meeting.  Judging by the almost-a-dozen folks who keep coming up to me and thanking me - even a week later - folks liked it too.  They all especially commented favorably on the poem with which I ended the talk.  One or two of our mature-in-the-gospel spiritual giants told me they were grateful to be seeing some things from a different perspective. 

It felt bad to have to tell them that ChatGPT wrote the poem for me, and other than a few edits to maybe 20% of the content, it was pure ai.

One guy even had me email him the poem, he liked it so much. 

How can something like new technology be so funny and so scary at the same time?

Link to comment
Share on other sites

11 hours ago, NeuroTypical said:

For the record, it didn't come up with the audio, that was added by a human.  AI lyrics are the audible version of those fingers. 

 

No, this isn't possible without some structure. I do not believe for a small moment that the AI had no template for determining structure. It had to be taught a template with elements like tempo, beat, rhythm, chord progressions, inversions, key relations and changes, and other musical stuff. You can't just take a collection of audio song files and expect an AI to imitate them without explaining what it is the AI is "listening to" any more than you could feed an AI a bunch of bitmaps of Rembrandt paintings and expect it to create a believable "new" Rembrandt knock-off without carefully explaining what it is the AI is "seeing".

Link to comment
Share on other sites

Way back when Apple introduced their first home computers, my teacher had me (along with a couple of other students) learn Apple Basic.  After we learned some fundamentals, she had us write all the commands that would be required to tell a robot to prepare a breakfast -- written in plain language format.  It was something like an introduction to pseudo code.  I don't remember the food that was required. But I remember one of the commands being to turn on the stove.  So, I'm guessing it was bacon and eggs.

When we compared notes, I had over 500 steps.  Others had 30 to 40.

I could have reduced my list to about 200 if I had used subroutines.  But subroutines hadn't been taught to me yet.  The reason mine was so much longer than theirs was because they simply assumed that the computer "knew" certain things like "how to turn on a stove."  While I only assumed that it "knew what a stove was."  Now we see that computers require a WHOLE lot more information about the world around them.  It doesn't even know what a stove is. 

This commercial shows they know virtually nothing about what a can or bottle of beer is.  They have to have every detail programmed into it.  Without that information that we humans simply "get" by looking at one in real life for the first time, it can't really "understand" anything.

Anything that can be defined by a mathematical pattern is easily programmed.  But anything that requires exact definitions (such as the size and shape of a can of beer) must be programmed in bit by bit.

What this reminds me about is when morfing was first introduced by the film Willow.  The person who developed it spend millions of dollars to create the software.  He had to program in EVERYthing at a fundamental programming level.

A few years later, a piece of morfing software sold for about $25.

Whoever actually takes the time to develop software that can describe the physical world around them (forget about esoteric stuff) in terms that computers can really understand will revolutionize AI.

But given that deepfakes have been around for over 20 years and are still easily detectable, it seems that the human face is quite difficult to program.  The face is not just a static object. It is a magnificent piece of bioengineering that changes with every fleeting emotion or inflection.  All of its geometric properties are almost always in flux.  A computer can't simply predict all those changes.  It has to have them programmed in.

It would have to have musculoskeletal structures defined.  Morfing of each muscle and the reaction of interacting muscles. Then the fluid dynamics of the skin layers and their effects on other parts of the face that are not directly linked to the muscles being flexed.  

I could go on.  But it would take a LOT for a computer to eventually understand the human face.

Edited by Carborendum
Link to comment
Share on other sites

1 hour ago, Carborendum said:

But it would take a LOT for a computer to eventually understand the human face.

The faces are pretty good.  Hands and fingers though… Oh my gosh are computers confused.

How hard is it to understand that there are 4 fingers and a thumb on each hand?

121AE5B7-9E87-4431-B15B-187210CADB20.thumb.jpeg.08c63132f92570a8f25dbfb35678ec9e.jpeg

Face is pretty good.  Hands!?

625CA597-A4D7-43BC-9E15-778ACB1B8A4C.jpeg.99010b9214b4e8a6e2ff422271de7874.jpeg

Edited by mikbone
Link to comment
Share on other sites

When I finished college (math and physics major) I went to work as a software engineer because I took a computer programing class and programing jobs paid better.  Early on there were no standard operating systems associated with computers and my initial work was done in machine code (first generation language) in which I became an expert.

As the industry advanced assemblers (assembly language which is the second-generation language).  Assembling a system to be downloaded and used was a project that could take as much as a week.  Updates were seldom.  When errors were discovered – programmers were required to create errata.   The essence of errata was to insert a machine code jump where the coding error exists.  The jump was to a common computer address where code to fix would be provided and at the end of the fix code would be a machine jump to the next address beyond the code in error.

The reason I am explaining this is because with the introduction of 3rd generation languages we started using compilers which made life in the programming world much easier – but during this era computer core was limited so efficient code was premium.  Good programmers could create more efficient code than any compiler.

Now days programming is done with 5th generation object-oriented languages.  Programmers capable of machine language or assembly code are for the most part non-existent.  

It is my opinion that AI is in essence a new generation of programming.  AI has taken rather odd turns from the AI I have created in the industrial workplace with industrial robots where the AI to operate a painting robot is quite different than a robot that builds pallets for shipping.   As smart as a painting robot may be with the most sophisticated AI available – it will never be able to build pallets for shipping.

I believe that as much as many are worried about losing their jobs to AI that there will be more professions created in connection with AI than will be replaced.  And the new jobs created will pay much better than the replaced jobs.  And yet there is still a caveat that even replaced jobs will reappear sometime in the future like the artwork expertise of a modern blacksmith.  It is the circle of history.

 

The Traveler

Link to comment
Share on other sites

5 minutes ago, Traveler said:

believe that as much as many are worried about losing their jobs to AI that there will be more professions created in connection with AI than will be replaced.  And the new jobs created will pay much better than the replaced jobs.  And yet there is still a caveat that even replaced jobs will reappear sometime in the future like the artwork expertise of a modern blacksmith.  It is the circle of history.

 AMEN. 
 

Technology is scary to those set in their ways or uncomfortable with change. But it helps to remember that tech almost always makes the world better for humanity. 

Link to comment
Share on other sites

3 hours ago, mikbone said:

The faces are pretty good.  Hands and fingers though… Oh my gosh are computers confused.

Apparently you didn't see the smiles.Smiles.JPG.c323007c1f9b6bfa1d573bac02b2b87b.JPG

That freaks me out like Pennywise.

Just watch it again looking for smiles throughout the video and tell me their faces look ok.

Edited by Carborendum
Link to comment
Share on other sites

Guest Godless
3 hours ago, mikbone said:

The faces are pretty good.  Hands and fingers though… Oh my gosh are computers confused.

How hard is it to understand that there are 4 fingers and a thumb on each hand?

121AE5B7-9E87-4431-B15B-187210CADB20.thumb.jpeg.08c63132f92570a8f25dbfb35678ec9e.jpeg

Face is pretty good.  Hands!?

625CA597-A4D7-43BC-9E15-778ACB1B8A4C.jpeg.99010b9214b4e8a6e2ff422271de7874.jpeg

This was my personal favorite. 

Screenshot_20230430_210730_Twitter.thumb.jpg.74a8950d89183e70d3217943e9127649.jpg

Link to comment
Share on other sites

25 minutes ago, Vort said:

ChatGPT will play you Zork on request.

Does anyone remember a text game like zork that dealt with a post nuclear war. I remember near the beginning having my character eat a can of food that was bulging and my character dying from food poisoning. I have been looking for this text game for years. It would have been around 1982. 

Link to comment
Share on other sites

2 hours ago, Emmanuel Goldstein said:

Does anyone remember a text game like zork that dealt with a post nuclear war. I remember near the beginning having my character eat a can of food that was bulging and my character dying from food poisoning. I have been looking for this text game for years. It would have been around 1982. 

Manny, was it from Infocom? The same company that made Zork, Lurking Horror, etc?

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
 Share