Let's talk about Artificial Intelligence

[old man emu]

Science Fiction has used Artificial Intelligence as a modern day bogeyman, but what that fiction predicts in the way of machines taking over personkind is hardly correct. I had to do some research to allow me to come up with that opinion, and what I found makes me think that the Pros well and truly outweigh the Cons. I can't see myself using it, but I'm not a Luddite who wishes to tear it down out of unproven fears.

 

Before we start, let's determine what Artificial Intelligence is. For one thing, it is not "intelligence" as we understand it in fish, birds and mammals. Basically the software that we use is a hunter/gatherer whose hunting ground is all the information present on the Internet. Added to its hunting skills is its ability to process what it has captured and serve it up to us in any number of formats that we desire. While AI can be used to produce intricate plan for tasks ranging from the simple to the complex there is one thing it cannot do. It cannot produce and original idea. It is incapable of abstract thought. 

 

Within the overall package that we call AI there are four separate AI domains. The term domain can refer to how resources are organized. In general, a domain is an area of control or a sphere of knowledge. Think of an AI domain as a area of activity such as Administration, Finance, Manufacturing or Sales within a business. Each domain has its own goals which may or may not overlap with other domains, but in general is independent of the others.

 

The domains of Artificial Intelligence are:

1. Generative AI, often referred to as Gen AI, is a domain within AI for the creation of new content, such as text, images, voice, video, and code, by learning from data patterns. Think Chat GPT.
2. Machine learning is a domain of AI in which computers autonomously learn and improve without being explicitly programmed.
3. Natural language processing is a domain of AI that deals with the ability of computer systems to understand and generate human language. 
4. Computer Vision is the domain of AI which allows computers to 'see' and comprehend the visual world.

Of the four domains, 1 and 3 are the ones with which the average person would have the greatest contact, unless operating a driverless vehicle, or conducting surveillance to spot undesirables in a crowd when 4 is the most useful.

 

So what are the Pros and Cons of domains 1 and 3? Domain 1 is the hunter/gatherer. It is used to trawl the Internet for information that provides an answer to a question posed by a human. The questions could be a simple request for a single piece of information, such as 'who was Peter Lalor?', to complex questions such as "produce a 200 word summary of peer reviewed papers relating to anti-cancer vaccines'. Domain 1 is capable of providing an answer in text which can reflect the style of a celebrity, or to suit the reading age of a target audience. Domain 3 can manipulate the text created by Domain 1 into formal or vernacular styles and/or into other languages.

 

Apart from the possibility of nefarious use of Domains 1 and 3, I see Domain 1 as a destroyer of originality. If a user becomes reliant on the output of AI then, while the output might be faultless, it is without personality. If you have read this far, and have read other examples of my serious writings, you will know that I have a particular style of constructing what I write. You will know that I often use uncommonly used words where perhaps I should have used more common ones. Domain 1 cannot reproduce my style, mainly because it has not been able to connect material I have posted with me. I'm told that all I have to do is direct the AI program to produce it response in the style of Old Man Emu, but I wonder if it could identify OME and OME's writings.

[rgmwa]

Hmmm. Sounds like something OME would write. It has his style, but is it really OME, or is it Chatgpt sounding like OME? 

[nomadpete]

1 hour ago, old man emu said:

makes me think that the Pros well and truly outweigh the Cons.

OME, don't you remember back when we were taught that Television will lead to great innovations to learning? And will usher in a new age of awareness and knowledge?

 

Now you are repeating that about a media that has all the same risks of misuse and misinformation?

[nomadpete]

Didn't you watch that YouTube link I posted?

 

The next stage of A.I. permits A.I. to do it's own research and also to 'adjust' it's own programming. To the extreme of priotorising it's own survival above that of humans?

[rgmwa]

I think AI will be fantastic for things like medical research but its development is likely to be driven far more by military imperatives and commercial profit than any altruistic peaceful purposes. It’s still early days and there is plenty of time for it to  have disastrous consequences as well as great benefits.

[spacesailor]

Does A I have a use in smart bombs  .

The old way , humans had to calculate the trajectory of the ' release to the impact .

Now they are released and the bomb adjusts it's trajectory to find that target by GPS.

The new A I smart bomb , uses internet image recognition,  to find the victim.

spacesailor

Edited Sunday at 09:59 AM by spacesailor
Lost a letter

[nomadpete]

Spacey, A.I. uses a camera to identify what a smart missile has been sent to hit. Then it guides the bomb to hit what it thinks it should.

 

In comparison,  what Russia is presently using is very old, out of date bombs that were left over after the cold war.

 

The A.I. guided thing does not need internet, nor gps, nor any connection to the sender. It is told what it's target looks like and goes looking for it.

 

 

Edited Sunday at 10:46 AM by nomadpete
final spellfix

[Marty_d]

What you describe OME is AI at the moment.  

People who are far smarter than me and work in the field of AI are extremely worried about the future.

On the other hand - you describe sci fi as making AI a bogeyman - try the Neal Asher "Polity" books.  Absolutely cracking reads.  They describe a future where AI develops well past human intelligence and then quietly takes over for humanity's own good.  Each planet is ruled by an AI which optimises every field of human/robot endeavour.  In fact medical tech progresses to the extent that unless you have to be scraped up in a bucket, an AI robot doc can probably save you - and if you have a memplant that records your memories, even that death is only a temporary inconvenience until you're either uploaded into a clone grown from your cells, or put into a crystalline brain in an android chassis.  Fascinating stuff.

[Jerry_Atrick]

4 hours ago, nomadpete said:

Didn't you watch that YouTube link I posted?

Can you pls repost, @nomadpete; for some reason, I can't find it.

 

 

In terms of threat to jobs, well, it depends what you mean. AI, like any other technology, is a tool for automation. Automation necessarily means replacing somethign a living thing does with a machine, usually cheaper and more reliably. In other words, automation disrupts the labour market. If we think of cars, if it weren't for the recreational and gambling aspect of horse riding, the horse may well have gone the way of the dodo. Noe, carriage makes, farriers, and those connected to the horsing industry, in relative terms is a small fraction of what it was. However, it was replaced by jobs in the automative industry from manufacturing, to selling, to maintaining, to parts, to customisations, etc. I haven't done the maths, but at least in absolute terms due to the decreasing real cost of owning cars and increasing population, the automotive industry employs many more people than the equine industry, and I am guessing that is also in real terms as the middle classes evolved.

 

The point I am trying to make is that as a disrtuptor, technology will only be a problem for the workforce as a whole if it doesn't create more jobs than it takes away, or there is sufficient capacity in the economy and its growth trajectory to take up the slack. In my days of programming, the more radical papers were suggesting my industry would render global unemployment in the 90% range. It was clearly a furphy for anyone in the industry, because it was more about efficiency than replacing jobs. And the other jobs that were created did not require a masters degree in maths and data and computational science to get work.

 

The AI boom, I think, though, is more problematic for the labour market than previous technology. Most techology was discrete in its application. It was more about optimising the work than replacing a human. As an example, I used to work on enterprise asset management systems in the nuclear sector. It was about helping the engineers, health physicists, techicians, bean counters, warehousing, and even buyers have the requisite information at their fingertips. We did dabble in early AI predicitive maintenance techniques, but most of the cerebral work was carried out by a person.

 

AI is very different. As an example, I put this prompt into Chat GPT:

"can you write me a stand alone java program that simulates an aircraft artificial horizon that banks a plane when using the left and right keyboard buttons and pitches the plane using the up and down arrow buttons"

 

Attached is the source code it wrote. ArtificialHorizonSimulator.java

 

If you have java installed on your machine, you can execute this by typing "java ArtificialHorizonSimulator", or on Windows you may be able to double click the file. There is no nasties in it, and you have the source to check, but run iut through your anti virus to be sure:  ArtificialHorizonSimulator.class(you will also need these in the same directory): ArtificialHorizonFrame$1.classArtificialHorizonFrame$ArtificialHorizonPanel.classArtificialHorizonSimulator.classArtificialHorizonFrame.class

 

It did have some foibles., the pitch line between the artifical sky and ground didn't move and when you banked, it was a square and didn't fill properly:

    Pitch is now > 40 degrees: (red is my addition)

 

Bank looks like this:

 

You can see it has a square background, so the edges aren't filled. But it took all of 2 minutes and even a qualified java graphics programmer would have taken a lot more time to get this writing it by hand.

 

So, I then asked chat GPT to moth the sky/land relative to pitch and to fill in the background properly, and then to add an altimeter and directional gryo and show relative indications:

 

It didn't quite get it right:

But with clearer prompting, and I guess the paid version. less than 5 minutes to provide this is a lot quicker than writing it by hand. And an experienced developer can take this, put the finishing touches on it, and voilla!

 

The thing about AI over more discrete automation is it can do a myriad of things. At work, we use AI to record meetingds, type up minutes and action points, put them in the Outlook tasks with follow ups. Seems remedial, but companies used to hire juniors to do that sort of thing.. not as many juniors are requied. The front office are now experimenting writing prospectuses and pitches for new products. In risk management and trading, we have been using AI for a while, reducing the need for the cerebral staff.

 

When quantum computing comes along, AI will be taken to a different dimension - not because of any groundbreaking new algorithmic advances, but sheer data processing output.

 

AI won't remove the need for the labour market, but it will be the biggest disruptor, and at the moment, I can;t see what will take the place of a significantly reduced demand for many of the people out there now. @Marty_d's work colleagues are right.. There will be a rise of skilled AI prompters; not because AI woll totally replace the workforce, but as it can ouput stuff far better and quicker than humans, it will fill a lot of roles that plug holes, leading to a significantly reduced demand for people.

 

 

 

 

[nomadpete]

Jerry,

I'm not worried by ChatGPT (etc).

 

 

Sure the series of youtoobs are probably largely Hennypenny (sky is falling), but in the past our fantastic tech advances have not realised their potential for good.

 

Seems that nefarious warmongers are better motivated than our educators.

[old man emu]

8 hours ago, Jerry_Atrick said:

In risk management and trading, we have been using AI for a while, reducing the need for the cerebral staff.

  It is that reduction in cerebral activity that I fear. I agree that AI is a useful tool for processing information in its many forms, but the information it sources is stuff that already exists - yesterday's news. That information is the foundation of what we have in the present, but AI cannot produce innovation of itself. It needs prompts from humans. In fact the word that is used to describe the requests a human makes to AI is "prompt", indicating that AI needs help to proceed in a desired direction.

 

I suppose I am being very restrictive in what I am talking about in relation to creativity. I'm just looking at it use in "literature". Could AI create the emotions in the writing that take a story from not much more than a news report to a timeless piece of prose? I doubt it because a great author of literature, or even music and art, puts heart and soul into the work, and AI has neither heart nor soul.