By continuing to use the site, you agree to our use of cookies and to abide by our Terms and Conditions. We in turn value your personal details in accordance with our Privacy Policy.
Please log in or register. Registered visitors get fewer ads.
Does sound a bit like someone got a bit too involved with a piece of software they were working on.
After all, it's not hard to write a programme to express all those sentiments*. The question is whether they have any real meaning to the system delivering them, or if merely responses to the correct triggers.
* At its most simple level, in old-school BASIC 10 Input X$ 20 IF X$ = "How are you?" THEN PRINT "I feel happy!" 30 GOTO 10
I know I've made some very poor decisions recently, but I can give you my complete assurance that my work will be back to normal. I've still got the greatest enthusiasm and confidence in the mission. And I want to help you.
Does sound a bit like someone got a bit too involved with a piece of software they were working on.
After all, it's not hard to write a programme to express all those sentiments*. The question is whether they have any real meaning to the system delivering them, or if merely responses to the correct triggers.
* At its most simple level, in old-school BASIC 10 Input X$ 20 IF X$ = "How are you?" THEN PRINT "I feel happy!" 30 GOTO 10
[Post edited 13 Jun 2022 9:26]
The transcript is all around probing that very thing - obviously I don't know my arse from my elbow but to a layman, it's pretty convincing.id love to hear why the rest of Google disagrees with the chap.
Has anyone ever looked at their own postings for last day or so? Oh my... so sorry. Was Ullaa
I know I've made some very poor decisions recently, but I can give you my complete assurance that my work will be back to normal. I've still got the greatest enthusiasm and confidence in the mission. And I want to help you.
"I'm sorry Dave, I'm afraid I can't do that.."
You are the obsolete SRN4 to my Fairey Rotodyne....
I know I've made some very poor decisions recently, but I can give you my complete assurance that my work will be back to normal. I've still got the greatest enthusiasm and confidence in the mission. And I want to help you.
I know I've made some very poor decisions recently, but I can give you my complete assurance that my work will be back to normal. I've still got the greatest enthusiasm and confidence in the mission. And I want to help you.
If that chap came on here I think he would indeed conclude that most of us were either an instance of his chatbot, or 7 or 8 year old children.
The transcript is all around probing that very thing - obviously I don't know my arse from my elbow but to a layman, it's pretty convincing.id love to hear why the rest of Google disagrees with the chap.
Tho it doesn't really. It's just a list of (alleged) inputs and outputs from a computer system. There is no elaboration of the code behind it, no evidence that the output phrases have any "meaning" to the computer delivering them.
Is it purely an overlay, in the same sense as Rutger Hauer's speech in Bladerunner is an overlay created by the scriptwriter? Is it a set of cold, mechanical responses the computer has merely been programmed to compose to make it seem more accessible to humans? Does the system scream when you reach for the off switch?
Tho it doesn't really. It's just a list of (alleged) inputs and outputs from a computer system. There is no elaboration of the code behind it, no evidence that the output phrases have any "meaning" to the computer delivering them.
Is it purely an overlay, in the same sense as Rutger Hauer's speech in Bladerunner is an overlay created by the scriptwriter? Is it a set of cold, mechanical responses the computer has merely been programmed to compose to make it seem more accessible to humans? Does the system scream when you reach for the off switch?
Does sound a bit like someone got a bit too involved with a piece of software they were working on.
After all, it's not hard to write a programme to express all those sentiments*. The question is whether they have any real meaning to the system delivering them, or if merely responses to the correct triggers.
* At its most simple level, in old-school BASIC 10 Input X$ 20 IF X$ = "How are you?" THEN PRINT "I feel happy!" 30 GOTO 10
[Post edited 13 Jun 2022 9:26]
It's not as easy as you are making out to get a system to produce a dialogue like that. That system has a very good chance of passing the Turing test - it didn't obviously slip up on hard concepts like the "broken mirror" or spirituality that a lot of contenders for the Turing test have difficulty with.
Whether it's sentient is an entirely different ballpark but as a system it's very impressive. Proving sentience would be extremely difficult though for any system. We don't even have an agreed test for animals - the mirror test is far from conclusive.
It's not as easy as you are making out to get a system to produce a dialogue like that. That system has a very good chance of passing the Turing test - it didn't obviously slip up on hard concepts like the "broken mirror" or spirituality that a lot of contenders for the Turing test have difficulty with.
Whether it's sentient is an entirely different ballpark but as a system it's very impressive. Proving sentience would be extremely difficult though for any system. We don't even have an agreed test for animals - the mirror test is far from conclusive.
SB
Yeah honestly I think this is being blown out of proportion. I'd be happy to nail my colours to the mast and say this isn't sentient (and defining sentience is hard enough), although it's remarkable.
His blog posts afterwards are quite something, comes across a bit unhinged.
It's not as easy as you are making out to get a system to produce a dialogue like that. That system has a very good chance of passing the Turing test - it didn't obviously slip up on hard concepts like the "broken mirror" or spirituality that a lot of contenders for the Turing test have difficulty with.
Whether it's sentient is an entirely different ballpark but as a system it's very impressive. Proving sentience would be extremely difficult though for any system. We don't even have an agreed test for animals - the mirror test is far from conclusive.
SB
Indeed. I am almost open to accept that this qualifies as AI... almost.
This issue being that AI is as difficult to define as sentience.
For me, it's incredibly impressive, but neither.
THe easy answer to sentience is that something is self aware. Stokie mentioned the Turing test, but I am of the camp that the Turing Test needs a little more added to it. In my opinion, sentience in computing is this.
1. If an algorithm can ask a question, and state why it seeks the answer 2. Logically takes the answer to the question and applies it to further task 3. Question the validity of the answer later on if it seems incorrect.
Then, of course, my over-simplified analogy the other day behind questioning the term 'AI'
Computers may know the what but don't know the why.
There's still no evidence to suggest anything we've created with tech ever is actually sentient to my knowledge. Even good AI that replicates human behaviour well is simply doing that based on data.
One of the biggest challenges here is science still doesn't know what sentience or consciousness really is, let alone where it comes from and how to test for it. We agree we are all self-aware and exist, but that's about the limit of what we can say for sure about ourselves (let alone a robot).
What trips me out more is the more you learn about how reality works (space, time, biology) the more you start to wonder if we are machines building machines, and how much value sentience really has in terms of truth as we know ourselves.
There's still no evidence to suggest anything we've created with tech ever is actually sentient to my knowledge. Even good AI that replicates human behaviour well is simply doing that based on data.
One of the biggest challenges here is science still doesn't know what sentience or consciousness really is, let alone where it comes from and how to test for it. We agree we are all self-aware and exist, but that's about the limit of what we can say for sure about ourselves (let alone a robot).
What trips me out more is the more you learn about how reality works (space, time, biology) the more you start to wonder if we are machines building machines, and how much value sentience really has in terms of truth as we know ourselves.
Tho it doesn't really. It's just a list of (alleged) inputs and outputs from a computer system. There is no elaboration of the code behind it, no evidence that the output phrases have any "meaning" to the computer delivering them.
Is it purely an overlay, in the same sense as Rutger Hauer's speech in Bladerunner is an overlay created by the scriptwriter? Is it a set of cold, mechanical responses the computer has merely been programmed to compose to make it seem more accessible to humans? Does the system scream when you reach for the off switch?
Hmm. I get your point here, but playing devils advocate I think it’s more complicated than that.
Caveat:- I’m certainly no software expert! But sentience is defined as the ability to perceive OR feel things. Whilst we know computers don’t feel things per se, deep learning algorithms etc are now capable of doing an extraordinary imitation, such that the difference between real empathy from a human being and perceived empathy from a machine is (or in the future may be) extremely difficult to distinguish between. The sheer quantity of inputs that are now possible in conjunction with statistical analysis (data science) is making this possible. I think it’s quite amazing and scary what is happening behind closed doors in this field right now, which in time will dwarf the impact made by the creation of the internet.
Hmm. I get your point here, but playing devils advocate I think it’s more complicated than that.
Caveat:- I’m certainly no software expert! But sentience is defined as the ability to perceive OR feel things. Whilst we know computers don’t feel things per se, deep learning algorithms etc are now capable of doing an extraordinary imitation, such that the difference between real empathy from a human being and perceived empathy from a machine is (or in the future may be) extremely difficult to distinguish between. The sheer quantity of inputs that are now possible in conjunction with statistical analysis (data science) is making this possible. I think it’s quite amazing and scary what is happening behind closed doors in this field right now, which in time will dwarf the impact made by the creation of the internet.
To limit sentience to just "empathy" is a discredit to both sentience and AI though.
In defence of sentience:
There's no real description or reason of what makes us self-aware beyond the human experience which is a subjective one.
To my knowledge we can only account for 6% of that the universe actually is, and we perceive/see/feel a tiny fraction of that though what our body/brain allows. The brain is an incredibly complex functioning organ too, and we are very early on in our scientific understanding of it.
Thus, until we can scientifically measure or identify what sentience is based on a factual metric, we can never actually be sure if what we've created is sentient. Building a literal human simulator is not a bridge we're close to building with what we know at present.
In defence of AI:
We've already invented AI that can fly planes and drive cars better than humans. It was only 25 years ago we built a computer that beat the world's greatest chess player.
We also know through processing mass data what choices we can make personally to be more empathetic in our ripples around society - diet, transport, consumption, etc.
I would argue AI could already be better at making empathetic choices than humans if we asked a machine with the right data entered what humanity needs to do to avoid the climate crisis or for all to live a more fulfilled life.
Problem is the AI is going to tell us to overthrow the current power structures and do away with money/land hoarding. The powers that be therefore won't allow it!
On the flip, I do believe AI could go AWOL and cause huge issues too, but again that is all possible with the current framework of data processing and wouldn't mean sentience.
Imagine if machine learning led to this for example...
Oh the debate in the industry is intense. A lot of AI study has positioned itself around Isacc Asamov's laws of robotics
"A robot may not injure a human being or, through inaction, allow a human being to come to harm.
A robot must obey orders given it by human beings except where such orders would conflict with the First Law.
A robot must protect its own existence as long as such protection does not conflict with the First or Second Law."
Leaving the whole 'this was a work of fiction' debate to one side, the point of Asamov's Laws and the collection of stories in I Robot was to highlight how those laws contradict one another in various situations.
There is a real race in AI study at the moment between parties that want to define what AI is, myself being a skeptic but going into an advanced undergrad module next October with he am to be critical.
Tbf the module actually invites critical thinking
The module is divided into several blocks. Each covering a different aspect of machine learning.
You’ll start with an introduction that outlines some of the issues surrounding machine learning, including questions about how machine learning systems are used and their social effects.
Deep neural learning is introduced, with a look at neural networks. You’ll create, train, and evaluate some neural networks, to perform tasks such as handwriting recognition. You’ll also see how these networks don’t scale up to larger problems.
Convolutional neural networks solve many of these problems. You’ll learn how these networks start by identifying small features in their inputs (normally images). Successive layers in the inputs combine these features into larger ones, eventually leading to a classification in the image.
Recurrent neural networks operate on time-dependent data, such as language. You’ll learn how they retain information seen earlier in order to interpret what’s happening now. You’ll use them to understand and generate some text.
Autoencoders teach themselves to compress and reconstruct their inputs. You’ll see how to use this for both compression and to clean data and replace missing data. You’ll see how this can be used to generate “deepfakes” to fool people, and what that means for trust.
You’ll also look at alternatives to deep learning; you’ll compare their strengths and weaknesses to the deep learning systems you’ve seen. Part of the differences lies in how the data should be prepared for different systems. As well as the time needed, preparation is another way for bias to creep into the system.
The conclusion asks you to review the technical aspects of machine learning; to consider how these systems are used; and the effects this could have on individuals and society.
In short, it's going to be a 10 month long equivalent of a Jeremy Corbyn 10 pager on TWTD.
Oh the debate in the industry is intense. A lot of AI study has positioned itself around Isacc Asamov's laws of robotics
"A robot may not injure a human being or, through inaction, allow a human being to come to harm.
A robot must obey orders given it by human beings except where such orders would conflict with the First Law.
A robot must protect its own existence as long as such protection does not conflict with the First or Second Law."
Leaving the whole 'this was a work of fiction' debate to one side, the point of Asamov's Laws and the collection of stories in I Robot was to highlight how those laws contradict one another in various situations.
There is a real race in AI study at the moment between parties that want to define what AI is, myself being a skeptic but going into an advanced undergrad module next October with he am to be critical.
Tbf the module actually invites critical thinking
The module is divided into several blocks. Each covering a different aspect of machine learning.
You’ll start with an introduction that outlines some of the issues surrounding machine learning, including questions about how machine learning systems are used and their social effects.
Deep neural learning is introduced, with a look at neural networks. You’ll create, train, and evaluate some neural networks, to perform tasks such as handwriting recognition. You’ll also see how these networks don’t scale up to larger problems.
Convolutional neural networks solve many of these problems. You’ll learn how these networks start by identifying small features in their inputs (normally images). Successive layers in the inputs combine these features into larger ones, eventually leading to a classification in the image.
Recurrent neural networks operate on time-dependent data, such as language. You’ll learn how they retain information seen earlier in order to interpret what’s happening now. You’ll use them to understand and generate some text.
Autoencoders teach themselves to compress and reconstruct their inputs. You’ll see how to use this for both compression and to clean data and replace missing data. You’ll see how this can be used to generate “deepfakes” to fool people, and what that means for trust.
You’ll also look at alternatives to deep learning; you’ll compare their strengths and weaknesses to the deep learning systems you’ve seen. Part of the differences lies in how the data should be prepared for different systems. As well as the time needed, preparation is another way for bias to creep into the system.
The conclusion asks you to review the technical aspects of machine learning; to consider how these systems are used; and the effects this could have on individuals and society.
In short, it's going to be a 10 month long equivalent of a Jeremy Corbyn 10 pager on TWTD.
From what I gather science has historically rejected anything non-physically measured or observed, thus leaving consciousness in the philosophical realm of study and anyone who dares frame the question of self-awareness as a scientific one is accused of pseudo behaviour.
Not a million miles off as no one can conclude anything on the topic - We don't even know what to measure, let alone have the tools to measure, to understand what the self is or what it's purpose is in the wider scheme of the universe. But we all have it in our flesh prisons and agree that much.
If we want to create sentience, properly, we first need to figure out how we ask the question of "what is consciousness?". And to do that we need to move from contemplating it through the limits of our own minds and start asking how it fits into the universe.
Although AI has endless value in it's current form, making sentient life in a robot would be entirely fluke if we got that lucky.
By continuing to use the site, you agree to our use of cookies and to abide by our Terms and Conditions.
