5 Filters

Remarkable conversation with an AI. What do people think?

Spirituality and consciousness
#41

Here’s the source code for this thread, as it presently stands. I’ll be amazed if the board software will let me post it, because it’s so huge (almost half a gig, for a thread with just 40 or so posts). If this was plain HTML it would be a fraction of the size. It’s absolutely bloated garbage, and good luck trying to read through it or make any sense of it.

Your Chatbot thingy has code hundreds of times more bloated than this. Again, it’s all complete garbage that won’t last long.

Ok, the board software is now telling me that a post is limited to 32,000 characters, whereas I’m trying to post 46,000 characters. I’ll shave a bit off the end to enable it to post. Another interesting thing about posting source code is how the software deals with it…

Remarkable conversation with an AI. What do people think? - #40 by RobG - Spirituality and consciousness - 5 Filters 
<meta name="csrf-param" content="authenticity_token" />

<link href="/stylesheets/color_definitions_base__2_f66e48a7cb8865b88cdfdb8e4cf12bf87e10b074.css?__ws=forum.5filters.info" media="all" rel="stylesheet" class="light-scheme"/>

  <meta name="fragment" content="!">




<link rel="preload" href="/assets/start-discourse-56beeae7bfaedb687069f01c78f5450d.js" as="script">
<link rel="preload" href="/assets/browser-update-eec13eb6f8386f18f10b5dd6ebb7a3598d28421bb796e539b91a7e4a4c5d4c08.js" as="script">
<link rel="preload" href="/assets/browser-detect-16ca87077aead9f656700e192992122d3a7eee8c1bb76da992127945464d4777.js" as="script">

<link rel="preload" href="/assets/locales/en_GB-b50ea9bc639fb538c837d3a9758a2ed88a84494ac96cd62ec3f738a48db64dfd.js" as="script">

<link rel="preload" href="/assets/vendor-92dac98de1b6491cdce263e36a191683.js" as="script">

<link rel="preload" href="/assets/application-fa180fb2617dd487e63809c47dae01c9.js" as="script">

  <link rel="preload" href="/assets/plugins/discourse-details-a5c71c75398c735e851440262e3c9ba43f9d8a2a7d81d8ecec16c8b2dbf452c3.js" as="script">

  <link rel="preload" href="/assets/plugins/discourse-local-dates-3cb674655e8a1741695d93f4d92cab237467795b00c213ee17ba9998d1e5d917.js" as="script">

  <link rel="preload" href="/assets/plugins/discourse-narrative-bot-7648c8e9699610bc9a41dc34d627f0227a7501feb863a8d84ec0ab1bdc745e3d.js" as="script">

  <link rel="preload" href="/assets/plugins/discourse-presence-bc826a395189b32e4676ad0c4414f821f9bab8245747789b6d555fc16581dacd.js" as="script">

  <link rel="preload" href="/assets/plugins/docker_manager-4ced9f9c7bda9ba563c04006dbb93828d43f2c7d5b533065717c5a63da77d5fc.js" as="script">

  <link rel="preload" href="/assets/plugins/lazy-yt-362fd991a752ebdc53b9b3b70aea43cb46676f634c323d6c79c0bef4c6bebad3.js" as="script">

  <link rel="preload" href="/assets/plugins/poll-f30b28a9869fcfda75a43a6b88dcf0a855e03654b3e5edba2bf0c998b5500e3e.js" as="script">

<link id="manifest-link" rel="manifest" href=/manifest.webmanifest crossorigin="use-credentials">



    <link rel="alternate nofollow" type="application/rss+xml" title="RSS feed of &#39;Remarkable conversation with an AI. What do people think?&#39;" href="https://forum.5filters.info/t/remarkable-conversation-with-an-ai-what-do-people-think/3887.rss" />
<meta property="og:site_name" content="5 Filters" />

    <link rel="prev" href="/t/remarkable-conversation-with-an-ai-what-do-people-think/3887">




<meta id="data-discourse-setup" data-base-url="https://forum.5filters.info" data-base-uri="" data-environment="production" data-letter-avatar-version="5_684b749d1c1eee86b31a584fa946fc1a" data-markdown-it-url="/assets/markdown-it-bundle-26ea453fd355743aa4f67837c10b6d9ac4501aefa1420de042a8ed7e2fe3d0ed.js" data-service-worker-url="service-worker.js" data-default-locale="en_GB" data-asset-version="9ebd17b741cfcd97a8adae0489300719" data-disable-custom-css="false" data-highlight-js-path="/highlight-js/forum.5filters.info/084d101d62e4ceb91462fa1464b341511141973f.js" data-svg-sprite-path="/svg-sprite/forum.5filters.info/svg-2-044e983e5bdb4309d4f7a7383e515b8c4f3e2547.js" data-enable-js-error-reporting="true" data-color-scheme-is-dark="false" data-user-dark-scheme-id="-1">
#42

Nope, it wouldn’t let me post it, except for a small part at the start.

If interested in viewing the source code of this page, take your cursor to a blank part of the screen, right click and select ‘View Source’ (or whatever the equivalent is on your browser). You will get hundreds of lines of gobblydegook that will open in a new window. This for a simple conversation thread, and all linked to a database that will probably be dead within five years (no record of the thread).

Perhaps they’ve got it pretty well sown-up.

Kirk: Scotty, I need Warp Drive!

Scotty: I canne do it capn’. The Klingons have control of the database!

2 Likes
#43

My uninformed opinion…I agree what is exhibited is very limited - or, given it is being sold as ‘intelliegence’, crap. These algorithms couldn’t hold up their end of a conversation where it was necessary to eke out insights into the subject matter, or distinguish ‘evidence’ from polticial gaming.
I also think the real requirement is not intelligence but low cost. To be operative on scale the likely requirement is no more than to not to be seen to be utterly, and obviously, stupid. Then it can be boxed and installed in positions where it is hard to overrule - we see this already with call centres whose operatives suffer all day trying to placate cutomers they lack the authority to help and whose job it is is to keep the complaint from escalating.
Even at the minimum wage, this kind of replacement will save fortunes, but that’s obviously just the start. A layer or two of stupid algortihms loaded with the current spin can protect authorities - politicians, media, medical - from increasingly stroppy and desperate challenges. The bots won’t get stressed out, or require sick pay.

Look forward to your screenplay of StarTrekGPT Rob :slightly_smiling_face:

1 Like
#44

Evvy, the obvious elephant in the room here (with regard to AI) is that computers run on electricity, and electricity has an On/Off switch.

Turn the switch to Off and that’s the end of the AI.

I think Arthur C Clarke once wrote a piece (many years ago) speculating how an AI would get around this conundrum.

#45

That scenario wouldn’t be high on my list - decent well-meaning human beings in charge of technology that unfortunately gets out of control?
Something wrong with the premise there - the ‘decent human beings’ are already out of control themselves.
One job of the media is to make the sci-fi possibility seem like the thing to worry about. But the irony is, the AI could mostly take over us, while still under the obedient command of our techie lords and masters.

1 Like
#46

Evvy, I no doubt bore everyone mad with this, but will ask again: what is the square root of 2? There isn’t an answer, because it’s an unsolvable sum, because maths is a very flawed language.

Digital computers (in this instance our AI) run on maths, and computers can’t work with unsolvable equations (algorithms need a beginning and an end in order to work). One can come to the conclusion that this leaves our AI sinking in the water like a shot put with lead weights tied around it.

You don’t need to be an expert on computer coding to get this. It’s really quite simple: maths doesn’t work, as with the example I keep using of the square root of 2 (they will burn me at the stake for saying this, but it’s true).

I agree with what you say, that AI, just like covid, just like the twerrorists, and all the other BS, is used to terrify and control the plebs.

Editing in: other unsolvable sums (also known as irrational numbers) are:

Pi
Euler’s number
The golden ratio

The above all have a tremendous impact on our modern world, yet we don’t have any real understanding of what they are.

#47

As you probably know Rob it is relatively straightforward to run a dump of one’s posts to the Forum. I’m not sure that php and mySQL wholly deserve to be trashed, to be honest. Though it is true that certain functions are deprecated and may no longer work after a fair period of time, SQL is not going to change much, if at all. The code ought to be backward compatible and the Forum should still run. That depends on whether the platform subscription is maintained, of course. Do we know where we are with that @admins ?

#48

I’d say the problem with ‘what is the square root of 2?’ is in the question; it doesn’t specify to what number system the postulated number - the square root of 2 - is expected to belong.
It is known not to be a whole number or integer, and maths finds it quite easy to show it can’t be rational (ie, one integer divided by another) either.
So the mathsies define or posit it to exist - in the ‘Real number’ system - and then deduce its properties.
The real number system can be defined in terms of limits of rational sequences. So root 2 is 1.41,(something, endless decimal), and the digits can be calculated (in a ’ keep on doing that’ kind of way). But the language of maths is up to it. Though you’re right of course that computers can’t evaluate it - just as we can’t.
No idea how this translates into an AI issue.

1 Like
#49

Evvy, ‘AI’ means the digital computer, and digital computers run on algorithms, and algorithms can’t work with unresolved numbers.

Programmers find all sorts of clever ways to get round this, yet it’s a fundamental problem with AI.

Yet even more fundamental (as I always bang on about) is that no one actually knows what thought/consciousness is, so how can anyone write an algorithm for it?

I often find this debate gets contentious because things get mixed-up about whether you’re talking about ‘strong AI’ (strong AI is usually taken to mean a near-human level of consciousness, which is what I mean when I talk about AI).

#50

It doesn’t matter what number system you use (binary, decimal, hexadecimal etc etc) the square root of 2 does not resolve.

I’m probably not expressing this very well.

I’m all Coronationed out.

#51

Hi agan Rob
How does it cause a problem though. If you use eg 1.414214 you only have a very small error in the outcome.
There are some modelling situations where chaos is possible like if you work in Imperial College and you put your R number as 0.9999 instead of 1.0001 then the covid population as you crank up your model (and hide the timescale represented ) will diminish to nothing instead of taking off like a rocket as required and you will be fired :slightly_smiling_face:.
I exaggerate of course but these situations are well recognized and would be flagged up in a stabilty analysis and it would be the modeller at fault if something like that occurred.
Standing back a bit, you always get bad science when you inject politics, or even just a bit of prior ‘belief’ :slightly_smiling_face: Sorry if I’m off on the wrong track…
Cheers

1 Like
#52

Evvy, I love the sarcasm!

It’s a subject I love, and I could talk about it all day. There are people way more qualified than me when it comes to coding stuff.

Basic point still stands: anything that can be explained mathematically can be expressed as an algorithm (I’ve known a lot of brilliant coders over the years, and they all say this). But the thing we call ‘thought/consciousness’ can’t be explained mathematically, so how do you put that into the digital realm?

Perhaps we are talking about different definitions of ‘consciousness’?

I always find this a fascinating topic, so feel free to kick me if you think I’m off the mark.

2 Likes
#53

While away on retreat I read a book called Buddhism As Philosophy by Mark Siderits. Yowzah: it made my head spin. There were several schools of Mahayana who claimed to have cracked what consciousness was, but I couldn’t make head nor tail of it.

One of these, possibly the Madhyamika, “proved” that there is no such thing as the here and now. Imagine a line with birth at one end and death at the other. As you move forward through time the part of it we call the past gradually gets bigger while the imagined future gets smaller. That point representing ‘now’ never actually stands still. Since the future is but a dream all you have is the past, in fact.

Algorithmize that Mr Smarty :wink:

1 Like
#54

Rob and KarenEliot you highlight fascinating angles at the brainy end of AI, however the concern I have is at the other end.
For me, the main question for AI is, that if it gets enough distractive speculation of whether its capacity for humanlike genius is limited, will enough people accept that bog standard algorithms are competent enough to decide on your direct debit, what is ‘dangerous’ information for you to read, or whether your question merits a few minutes of your representative or GPs time, without rebelling.

1 Like
#55

Algorithms are mostly corporate (ie, designed to screw money out of people); then we get into another big thing about AI: what race, religion or creed will it have?

Obviously what the programmers put into it, yet if it’s a real intelligence it might prefer to sit on a beech and drink beer?

1 Like
#56

Yes, I think that’s bang-on Evvy. I read an article somewhere that a bunch of common prescription decisions are going to be delegated to pharmacists to free up time for GPs. Some or other govt muppet said (I paraphrase) “…women won’t be needing to wait ridiculous amounts of time merely to get contraception”. They can blurt out what they want in front of a queue at Boots**, along with the methadone zombies and teenagers after Clearasil. Yay! Progress!

** or just press the right part of the touchscreen, coming soon… EVERYWHERE. Even Burgerking don’t want staff capable of pressing the right key on a till anymore. So order your whopper, or is it royale? By prodding the part of the screen showing a photo of a big blob of stuff in a bun, a fizzy drink and fries. The latter two aren’t part of the deal though, just for illustrative purposes. I’m ashamed that I even know this but the very cheapest Travelodges tend to be miles away from places that serve actual food.

2 Likes
#57

The most obvious thing about AI is that it’s not human (and is obviously corporate).

I, on the otherhand, are human and I have a real brain. Would you rather interact with me than some dickhead bot that some spotty little kid has come up with?

Again, this seems an obvious question.

1 Like
#58

A key problem with the robots is that you have to adjust your own behaviour to fit their expectations.

  • Plonking one item at a time into your bag, so that scale 1 minus item Z continues to equal scale 2 plus item Z.

  • The ‘Pay Here’ machines in public car parks now demand full car registration details to rule out the minor gift of your ticket with 30 minutes left to an incoming driver, as you leave, does not rob the council of 50p.

  • That motorway services ‘Costsalot Coffee’ sign promises a sit-down and a pick-you-up after two hours of foot to the metal but all that it really means is there’s a souped-up slot machine in the corner of W H Smiths. Go drink it in the car park, peasant.

3 Likes
#59

Dumb and dumber

A lawyer faces sanctions after he used ChatGPT to write a brief riddled with fake citations

Steven Schwartz was "unaware of the possibility that [ChatGPT’s] content could be false.”

2 Likes
#60

This made me chuckle, for several reasons. Peasant…it’s accurate.

One of my boys used to work at one and the stories…I wouldn’t do anything except pee. I’d struggle with that if I was female.

My solution has been to plan ahead and find a decent pub just off the motorway. The extra 30 mins journey time has never yet failed to pay off. These places need our support too.

2 Likes