State of play 2023

[LinuxGam]

There is very little decency left in a some affiliates, they change what ever they want in their terms or only slightly worse, just adjust figures. As we mostly all accept ( this didn't happen in the old days), that they can change the terms of their agreement at any time (totally). This would not be acceptable in any lands based agreement/legal financial agreement.

On a slightly different route, after believing Google had massively changed the world of search (mostly for the good, They even said so!) This has changed as much as people predicted, it's now a cash grab for paid spots in any terms monetizatible. Their search now is not good for users and doesn't encourage people to supply free content.

While I have nothing against AI and in fact like it, a lot of the full text AI, is based on other people's historical responses or websites, or copyrighted art and even logos. They have taken the time to really investigate a subject and then having their answers re-written by AI by websites that have no intention of spending this much time (or any, its fully automated) is a sad place to be.

There is also the old school issue, where the bigger sites/corps buy up all the sites they think are valuable and this makes it harder and harder to compete. Basically you end up trying to out produce content, web design, coding, advertising, even respect in search and hosting servers with limited resources.

I think the latter point is capitalism, but for me the bigger concern in the industry is the way search seems to work now, giving exponential favor to these sites/networks that can also out pay everyone in search. However, I have no issue with this, as everyone like this will fall at some point when something new and better comes long.

Love to hear your comments.

[TheGooner]

The answer to circumventing googles practice is to use a different search engine.
Be part of changing search engine ethics.

I recommend DuckDuckGO. -> https://duckduckgo.com/
And if you like the results then spread the word.

[universal4]

While I agree that at time it seems there is no decency left, I want to believe that those who continue to work in more ethical manner will continue to exist because they are more honorable and those that are less affiliate friendly will have a more difficult time and hopefully will lose more market share due to some having more friendly terms. (never lose hope)

As for google, I think many see that their results are not as accurate as they used to be and sadly it doesn't look like any competition will change that, but with AI improvements it is still possible.

As for AI, hopefully AI itself will become more intelligent and be able to determine what content is AI generated and what is human generated so that the human content can rank better. It is a slippery slope and only more time and maturation will determine that I think.

Agree that some of the larger brands will likely continue to gobble up what they can when they can.

Trying to stay more optimistic with a general sense of perseverance and not giving up. But having some levels of pessimism and being reminded to keep up with both sides of the discussion is never a bad thing.

Rick
Universal4

[LinuxGam]

I'm pretty sure DDG results are BING API results repurposed.

That doesn't make them wrong, but at least they promise your privacy.

Ive never liked BING results even though I can get a page in their bordered box at top, it's not a huge change in revenue... thought it helps when they both kill me.

[Sherlock]

As for AI, hopefully AI itself will become more intelligent and be able to determine what content is AI generated and what is human generated so that the human content can rank better.

The AI(1) now is being trained by other AI(2). The text is served to us when AI (2) can no more recognize that the text by AI(1) is computer produced.

So in other words we can see only the text that is "good enough". The bad enough texts are being dumped or rather they are not even written by AI, because AI knows that they could be marked by other AI as AI product.

So in other words asking AI to produce that bad texts, so other AI can see it, is pointless, because the core of AI learning is exactly the opposite.

The AI itself will become more intelligent and (therefore) it will just write even better texts, because it will bee more intelligent, that is it.

This is not a whack a mole game, where some superior AI on a good side will keep the inferior AI on leash. AI(1)=AI(2) Anything that "knows" AI(1) knows AI(2) and vice versa.

Somehow I feel that in the end it will bee all about access to sources/energy/computing power/money as always. More centralisation, more monopoly crony capitalism, less influence of state. Combine it with crypto and you get just a classic dystopia from Neal Stephenson. Nothing unexpected.

[LinuxGam]

The AI(1) now is being trained by other AI(2). The text is served to us when AI (2) can no more recognize that the text by AI(1) is computer produced.

So in other words we can see only the text that is "good enough". The bad enough texts are being dumped or rather they are not even written by AI, because AI knows that they could be marked by other AI as AI product.

So in other words asking AI to produce that bad texts, so other AI can see it, is pointless, because the core of AI learning is exactly the opposite.

The AI itself will become more intelligent and (therefore) it will just write even better texts, because it will bee more intelligent, that is it.

This is not a whack a mole game, where some superior AI on a good side will keep the inferior AI on leash. AI(1)=AI(2) Anything that "knows" AI(1) knows AI(2) and vice versa.

Somehow I feel that in the end it will bee all about access to sources/energy/computing power/money as always. More centralisation, more monopoly crony capitalism, less influence of state. Combine it with crypto and you get just a classic dystopia from Neal Stephenson. Nothing unexpected.

There is also the very real likelihood AI will become Chinese whispers. Right now, it learns based on the huge amount of information people have worked hard to learn and research.

As it then learns on AI's interpretation of others work and less and less people are willing to publish hard worked research or articles as they are drowned out by AI crap and Google. It learns from other AI un-edited sites or even edited.

Before long, AI is learning from slightly or very wrong AI and it gets to the point nothing is actually true.

I would love to say they have the intake of social media to keep things real... But all the most popular post I see are peoples opinions on politicians/celebs and often have no proof or worse.

AI as it is right now, needs to learn from correct or close to correct data.

[Sherlock]

Is not then the AI just doing the work that was done by social networks? Amplifying the dull issues, to which most people devote their time.

Is not your declamation exactly the political statement that you apparently do not like?

Who are we, being sure that the celebrity hunt and political debates, do not have some deeper sense?

I do not personally care about social networks (GPWA is my last gate where I talk with people who I do not know) or AI or politics. But the nearly 10B of people need some valve how to waste their lives, potential and money. We are collecting those money after all from the most desperate from desperates.

There was zero information on social networks before and people were stupid from Google even before. The trend was and is clear: the majority of society is refusing to behave as adults and the majority resigned to achieve something like human development. At last majority of the 10% minority of the world population called "The West".

There is no significant difference to talk to people on the internet (or lately even physically, because internet is the reality) and to some AI. Many years ago, sometimes around the time when Skype pushed away ICQ, I started to be sure that even people do not pass the Turing test anymore. I could give endless examples, how people do not understand each other, just because they do not want to understand and interact with each other. I can see and feel just their egos and traumas shouting at each other. So... I refuse to say this is wrong anymore. AI will give them endless opportunities to create their own hell, because they ask for it.

The society is dying (to be reborn) and if AI will bring the coup de grace, so be it.

It is really not set in stone, that each life has to be meaningful, or that society must have some aiming or higher goals. It is just political statements saying so, because nobody wants to vote for politicians with depressive news, who will tell to voters, that politics or social networks is just something infinitely less important than anything else.

The people are and will be divided to majority that will eat more and more xanax and who will see more and more how trapped they are through AI directed series on Netflix, which will just make them more depressive, because they will always have some cool excuse why they can not do exactly what they need to do. And to the minority that will keep something human through the dark ages.

What I want to say is that majority is not important for survival. People can keep on being idiots. Evolution never went through the majority. Majority always died out.

If you know that AI is wrong, social networks are wrong, politicians are wrong and voters/consumers are always manipulated, then there is only one choice. To refuse that world or to be part of it with random pretext (that will be confirmed by millions of others on Facebook, so it will not be so hard to believe in it - just the friction between that belief and reality will cause you some mental problems, that will bee called by doctors "a disease", so you can spend rest of your life by curing it, while no remedy [in a form of chemical substance or therapy] will exist).

[LinuxGam]

Is not then the AI just doing the work that was done by social networks? Amplifying the dull issues, to which most people devote their time.

Is not your declamation exactly the political statement that you apparently do not like?

Who are we, being sure that the celebrity hunt and political debates, do not have some deeper sense?

I do not personally care about social networks (GPWA is my last gate where I talk with people who I do not know) or AI or politics. But the nearly 10B of people need some valve how to waste their lives, potential and money. We are collecting those money after all from the most desperate from desperates.

There was zero information on social networks before and people were stupid from Google even before. The trend was and is clear: the majority of society is refusing to behave as adults and the majority resigned to achieve something like human development. At last majority of the 10% minority of the world population called "The West".

There is no significant difference to talk to people on the internet (or lately even physically, because internet is the reality) and to some AI. Many years ago, sometimes around the time when Skype pushed away ICQ, I started to be sure that even people do not pass the Turing test anymore. I could give endless examples, how people do not understand each other, just because they do not want to understand and interact with each other. I can see and feel just their egos and traumas shouting at each other. So... I refuse to say this is wrong anymore. AI will give them endless opportunities to create their own hell, because they ask for it.

The society is dying (to be reborn) and if AI will bring the coup de grace, so be it.

It is really not set in stone, that each life has to be meaningful, or that society must have some aiming or higher goals. It is just political statements saying so, because nobody wants to vote for politicians with depressive news, who will tell to voters, that politics or social networks is just something infinitely less important than anything else.

The people are and will be divided to majority that will eat more and more xanax and who will see more and more how trapped they are through AI directed series on Netflix, which will just make them more depressive, because they will always have some cool excuse why they can not do exactly what they need to do. And to the minority that will keep something human through the dark ages.

What I want to say is that majority is not important for survival. People can keep on being idiots. Evolution never went through the majority. Majority always died out.

If you know that AI is wrong, social networks are wrong, politicians are wrong and voters/consumers are always manipulated, then there is only one choice. To refuse that world or to be part of it with random pretext (that will be confirmed by millions of others on Facebook, so it will not be so hard to believe in it - just the friction between that belief and reality will cause you some mental problems, that will bee called by doctors "a disease", so you can spend rest of your life by curing it, while no remedy [in a form of chemical substance or therapy] will exist).

That's a pretty deep and heart felt post. I appreciate your time. I have no issues with AI, but I know how it works having trained various Neural Networks myself. If you learnt for other data and that data is wrong you then produce results based on that data. Yet people keep publishing AI content and eventually it will learn from a skewed reality.

I LOVE AI.... But not to reword peoples hard work. The ironic thing, is that barely affects me.. Just looking at whats happening.

[TheGooner]

I don't believe in AI - or at least I don't believe that the current things we have like ChatGPT are actually AI or even close.

They're very clever "expert systems" with good language parsing, so that they can produce good words and regurgitate facts and data nuggets very quickly to all types of people with very little technical knowledge required.

It IS quite useful. Today I typed in one sentence into ChatGPT asking for some blurb for an ETSY shop I have just been opening - I said give me a few paragraphs from a gruff old orcish merchant opening his shop. And it instantly popped out with some prose that I could re-purprose quickly.

It took me five minutes to do and I like it - if I'd crafted it myself then it would have taken and hour - and I'd still be struggling with it. But was that bot intelligent? No. It broke down my command according to instructions (Open Shop, Selling Items, style old Orc) referenced it's style catalog, and probably ripped off some text from somewhere in the web before slightly altering it.

ETSY Shop -> https://www.etsy.com/nz/shop/CollectablesByPaul (Selling Vintage Hand-Painted WarHammer Citadel Miniatures)

"Paul's Emporium of Wondrous Items is opening the doors to its newest shipment. Let me tell ya, it's filled to the brim with loot unlike anything you've seen!"

That's not intelligence. It's very clever code. But those two are very different things.

Similarly, I created the Green Guys on the shop front with a drawing AI, and on the first attempt, but again it's just a clever language parser sitting in front of a drawing package that rips off exiting images and slightly alters them.

It's a great new front end for non-artists like me who don't have the patience to learn art packages or to tweak knobs, and apply light, saturate colours etc, but again it's not intelligence it's an expert system following code and instruction sets.

-----

What would make true AI for me?

AI would actually learn from each interaction with end users (not just the carefully processed data nuggets it's creators have created).
AI would need to have wants and desires and have actions that it wanted to do.
AI would be pro-active, it would prompt me and want to have particular information or actions done.
AI would have a coherent understanding of the physical world, concepts of things, and probably have audio and visual senses too.

These ChatBots, DrawBots have none of these things, I can "feel" how they are written and created. They ARE very clever programs, but in no way are they intelligent. The output is well constructed language that COULD have come from intelligence - but it didn't - it was formed by a very proficient language interface. And THAT fools most people as good language equals intelligence.

These bots are not even trying to learn from their interactions, they have no "desire" for knowledge, or continuance and no concept of self. They have been constructed in a way that probably makes that impossible.

They are a small step on the path to AI - but possibly a step down a dead end path. The language manipulation algos will be useful - but not the base code that source nuggets and regurgitates them.

[universal4]

This discussion has been nteresting and thanks for your take on it gooner. I agree with a few of your points

These bots are not even trying to learn from their interactions, they have no "desire" for knowledge,

Agree, they only have a desire (programmed in to their backend) for DATA which is far different than actual knowledge. As many have said, it doesn't matter if the data is accurate, just that it exists and the more incorrect data it finds, the more it presents the incorrect data as fact.

I do think the artwork creation tools can be helpful, I have not spent much time with them but may in the future.

Rick
Universal4

/side note: Cool collectables and whatever you used to create that piece of artwork, very cool...by the way is that green guy your alter ego?

[LinuxGam]

I can’t say wether it’s intelligent or not…. I get your point on that.

I can absolutely assure you though their is ZERO code written that makes options or choices….

Having written my own Neural Networks and put through football stats and various other tests etc, there is absolutely no code/parsing…. Other than making sure the input data is normalised and in the correct format and designing the format of the neural network.

I have also downloaded the smaller predesessors to ChatGPT and the Dall-e and ran them on my 24GB 3090GTX.. again there is zero code or parsing…

I’t is a huge network of neurons thats are trained with weights using a lot of data and they make decisions what paths to make, very similar how our brains works.

There is no written logic or code….. only enough to get the data in and read the answer…

It’s why people refer to them as blackboxes as even the people who trained them with the data can’t say exactly how they work…..

[LinuxGam]

I took a list of every name and surname i could find in the world and train a custom neural network. Every single output I got was a name that sounded 100% plausible, yet didn’t exist. I generated 100000 of unique “fantasy” names and every one read like a book author created it rather than it being “adfqwfwefqwde” it was a name a surname that sounded real but almost always unique.

I also had some reasonable success with football stats and results… which is something I will go back to when I have more time.

Photo recognition is not some clever trick with parsers…. Or if/then logic… The photo is split from a 3D array of dots into and array of numbers representing the colours at each location…. The network is then fed thousands of different images and can accurately guess animals and as we know even do facial recognition with amazing accuracy. There is no if dot this then this…logic written by someone.

This is a complex and large subject but it not too hard to get started with Python if you want to play.

[TheGooner]

Photo recognition is not some clever trick with parsers…. Or if/then logic…

Based on your code that you've produced for livescore stuff I know that you know that there is a lot more to code than "if/then" logic.


Based on facial recognition programs (that I have some banking experience with) I can tell you that you can describe a persons face in a short series of numbers that creates and index into a database of similar faces.

Outline the face, measure the distance between eyes to nose to mouth, adjust for angle, colourise hair and skin and you get somewhere between 5-10 category numbers that will give you close matches - often unique hits.

Guessing animals? Childs play - if you have the software and database. It's not "if then" - it's pixel analysis, content and contrast ratings, and basic geometric measurements. But it is basic code.

Very cool, and quite amazing, but nothing more than image manipulation and colour / ratio percentages and code (that does not learn).

[LinuxGam]

Based on your code that you've produced for livescore stuff I know that you know that there is a lot more to code than "if/then" logic.


Based on facial recognition programs (that I have some banking experience with) I can tell you that you can describe a persons face in a short series of numbers that creates and index into a database of similar faces.

Outline the face, measure the distance between eyes to nose to mouth, adjust for angle, colourise hair and skin and you get somewhere between 5-10 category numbers that will give you close matches - often unique hits.

Guessing animals? Childs play - if you have the software and database. It's not "if then" - it's pixel analysis, content and contrast ratings, and basic geometric measurements. But it is basic code.

Very cool, and quite amazing, but nothing more than image manipulation and colour / ratio percentages and code (that does not learn).

Thats prob how they used to do it

I have downloaded many neural models and they are literally just a HUGE array of weights and i can run them through exactly the same software as I did my own trained ones.. therefore I know there is no logic… You can even generate Picasso images or a giraffe riding a bear etc…

[LinuxGam]

You could very easily write software to win/draw naughts and crosses very easily… You can’t write logic to beat a player at GO purely with logic as the combinations run into numbers unimaginable in our universe.

Yet they have now train AI to beat the world champion… and they trained it first using 10000000s of games then ass it got better it they trained it by playing against itself… Thats and amazing and scary thing…

With game playing there sometimes is some logic to prevent illegal moves etc… but the decisions are made in the black box and know one knows exactly how.

Watch the video, prob on YouTube of the World Champion being beaten… The AI was making moves that made no sense to the world champion or the commentators… So everyone was like.. whats it doing?

[TheGooner]

I have downloaded many neural models and they are literally just a HUGE array of weights and i can run them through exactly the same software as I did my own trained ones.. therefore I know there is no logic…

Of course there is logic - you just don't get access to it or see it.

What use are "weights" unless there is a core system to interpret the weights and numbers and apply meaning, direction and logic to them.
The logic is there to do that, and to do a million other things in the event of weight mismatch, number overload or other unexpected events.

Neural nets are useless without the core programming to interpret the data, and that programming is complex.

------------

As for the learning machines you've described with Chess and Go, these are very different from the ChatGPT and other public "AI" because they DO learn from all their previous attempts - whereas ChatGPT et. al. don't.

These machines do at least improve and build new strategies not explored by humans, but by their very nature they are specific task based engines that have no knowledge or use outside their specific game parameters.

-------------

Look don't get me wrong - I am ready and waiting to welcome our new AI overlords.
I'll be the first in line to purchase a new AI, feed it power, teach it, and live with it at home.

But we're a long way from that point - despite the hype, PR, and mass media hysteria.

[universal4]

Part of the problem with any results being "weighted" is that flawed or incorrect data will lead to incorrect results due to the "weighting" not be correct based on such flawed data.

It also opens the door to "man" also affecting the "weight" to the benefit of their agenda toward a desired outcome they choose.

Rick
Universal4