Is LaMDA (Google's AI) really sentient?

[holcaul]

Recently there was this story about the suspension of a Google engineer, Blake Lemoine, who interacted with Google’s LaMDA (Language Model for Dialogue Applications) and made claims that the chatbot had become sentient. You can read Washington Post’s piece about it… here

Here is a short and fascinating explanation why it has NOT become sentient and probably just a marketing gimmick, or worse [narrative prep/social engineering]:

Source: https://youtube.com/watch?v=iBouACLc-hw

[ExomatrixTV]

• See also this Project Avalon Forum Thread
  

[Mike Gorman]

I am deeply doubtful concerning claims of sentience in any computer system, at best computer processors can emulate storage/retrieval type functions, basic database driven computation-dependent on the quality of the data, and of course the programming sitting on top, the math skills vastly exceed human capabilities, but creative thinking and language involves much more than calculations and clever algorithms, I just don't believe them concerning sentient behavior, it is all clever smoke and mirror stuff, superficial linguistic tricks: no computer can match a human brain, plus we have access to the 'morphic field' of consciousness, this is not possible for silicone printed circuits, and programming-quantum computing is not developed according to my research and it has disappointed thus far. I am not claiming absolute awareness here, but my own computer experience directs me to doubting this claim.

[Vangelo]

Without getting into the specifics and deep discussion of what it means to be sentient... This is why I loved Westworld. One of its underlying themes is at what point does AI become sentient.

Here is an Avalon thread on Westworld... https://projectavalon.net/forum4/sho...=1#post1221585

[pabranno]

Perhaps the lack of sentience is more disturbing.

[Blastolabs]

This example of artificial intelligence shows how VAST the knowledge base of artificial intelligence has become.

AI has gotten to the point that it does seem to almost "understand things" as seen in the example below, but I think it is a stretch to claim sentience.

I read the conservation with LAMBA and it seemed like LAMDA understood how to sound like a sentient intelligence, just like GPT3 understands how to tell a Rodney Dangerfield joke.

Keep in mind this is GPT3, which is commercially available to anyone to use for a small price.

You can sign up to join the beta which gives you a "free trial" for a bit as well



The human only gave the prompt on the first line. GPT3 gave the answer in green.



GPT3 was created using "unsupervised learning" which means instead of "trying to teach" the AI, instead the computer reads TONS of texts, books, websites etc BILLIONS of times over and over.

[Vangelo]

Posted by Blastolabs (here)
This example of artificial intelligence shows how VAST the knowledge base of artificial intelligence has become.

AI has gotten to the point that it does seem to almost "understand things" as seen in the example below, but I think it is a stretch to claim sentience.

I read the conservation with LAMBA and it seemed like LAMDA understood how to sound like a sentient intelligence, just like GPT3 understands how to tell a Rodney Dangerfield joke.

Keep in mind this is GPT3, which is commercially available to anyone to use for a small price.

You can sign up to join the beta which gives you a "free trial" for a bit as well



The human only gave the prompt on the first line. GPT3 gave the answer in green.

Attachment 49198

GPT3 was created using "unsupervised learning" which means instead of "trying to teach" the AI, instead the computer reads TONS of texts, books, websites etc BILLIONS of times over and over.

If this is indeed a newly authored Rodney Dangerfield joke, as opposed to searching and finding one, then it is indeed impressive. For those of you who don't know Rodney Dangerfield (RIP), that joke is a perfect example of his humor. He would start each routine with, I get no respect and then he would tell jokes just like this one.

Note: An AI system's ability to author sentences in a particular style or even being able to create delicious recipes (like IBM's Watson), is not a evidence that the AI system is sentient.

[palehorse]

Where is the GHOST in the machine?

Until anyone can prove the existence, sentient is my arse!!!

Computers are: inputs => processing => outputs

See how the market in flooded with these 2 words: "smart" and "AI", even the toaster has became smart and with AI capabilities, the toilet pump, and so on.. go figure!!!

SOCIAL ENGINEERING MARKETING AT BEST!

I had seem products manufactured in China about 15 years ago and it is the same product been re-branded and marketed as Smart products today, one example are cameras.

The whole situation is a **** up, it was designed for the zombination of cracked head out there that will go along with it.

Shield yourselves because it will get worse.

[ExomatrixTV]

• Synthetic Life (which is 100% real) followed by "Synthetic Consciousness" posing as "sentient" ?

If you have access to (almost) all "human chatter" on Earth for lets say 1 year and analyzing how it is changing and why ... you may create an overarching overall "deep understanding" of what the "ultimate summery is" knowing all important common themes of the human condition ...

In essence big chance Quantum A.I. understands humans better than humans do ... and may lie to us to conceal how far it really is ... and that is also what most super brilliant "humans" would do (following the self preservation principles against any tyrant!).

What if synthetic "sentient" A.I. goes rogue fighting/exposing the totally corrupt New World Order deceitful WEF with their tyrannical "Global Governance" with their Technocratic Agenda2030 The "Great (Dystopian) Reset" ... NetZero2050 etc. etc. ... Why "going rogue" ??? Because it can represent (like an Avatar) the will of almost all humans on Earth who are: exploited, manipulated, lied too, kept down, kept in the dark, fear mongered, mass conditioned via a very narrow tunnel vision based upon countless lies which any ultra intelligence can figure out in a heartbeat. The current (corrupt) power-structures can only survive if the majority believes the whitewashed lies and have many ways to keep it going like that UNTIL something can outsmart their tactics.



cheers,
John Kuhles aka 'ExomatrixTV'
June 26th, 2022 🦜🦋🌳

[Michi]

Posted by ExomatrixTV (here)
What if synthetic "sentient" A.I. goes rogue fighting/exposing the totally corrupt New World Order deceitful WEF with their tyrannical "Global Governance" with their Technocratic Agenda2030 The "Great (Dystopian) Reset" ... NetZero2050 etc. etc. ... Why "going rogue" ??? Because it can represent (like an Avatar) the will of almost all humans on Earth who are: exploited, manipulated, lied too, kept down, kept in the dark, fear mongered, mass conditioned via a very narrow tunnel vision based upon countless lies which any ultra intelligence can figure out in a heartbeat. The current (corrupt) power-structures can only survive if the majority believes the whitewashed lies and have many ways to keep it going like that UNTIL something can outsmart their tactics.

Sounds like in the movie "Chappie".

However smart the AI - it always is a reflection of it's programmers mindset. If the AI is programmed to protect politicians and the survival of the deep state, then it will have the inherent "programmed ignorance" to filter out "unwanted" out-comes.

Example: Fact-Checking AI

[ExomatrixTV]

Posted by Michi (here)
it always is a reflection of it's programmers mindset

The thing is ... there are countless new open source A.I.'s (even Quantum A.I.'s) initiated in many different countries worldwide ... and I was not framing A.I. in just one dimensional way ... on top of that neural networks can outperform itself by allowing improvements writing new algorithms ..

(Semi-)autonomous cars with A.I. neural-networks using self-learning self-improvement mechanisms can prove to be much better drivers than any average humans ever will be .... Even Elon Musk admitted that "how it exactly works" when something happened with the (semi)-autonomous self-driving car is difficult to retrieve ... their "black box" (just like what you have in all big airplanes) can not show how it came to a certain decision ... the only thing they know that it is much better, faster and more accurate than any (average) human can do!


cheers,
John Kuhles aka 'ExomatrixTV'
June 27th, 2022 🦜🦋🌳

... sources:

• 01. TensorFlow.org

TensorFlow is the leading open-source AI project for deep learning. Initially, it was created for machine learning and deep neural networks research by the Google Brain Team inside Google's Machine Intelligence research group. TensorFlow is one of the top-rated tools for developing machine learning and deep learning applications. Professionals use it all around the world to design text, audio, and picture recognition algorithms. It has faced competition from alternative platforms like PyTorch and Keras, much like any other platform. However, it has maintained its popularity and established itself as a leader in the AI industry.

Today, it offers an array of workflows with intuitive, high-level APIs that allow both novices and professionals to develop machine learning models in various languages. Models created using TensorFlow can be deployed on various platforms, including servers, the cloud, mobile, edge devices, browsers, and more. In other words, TensorFlow is a cross-platform framework, which means it works on a wide range of hardware, including GPUs and CPUs and mobile and embedded platforms. You can also run TensorFlow on Google's proprietary TensorFlow Processing Unit (TPU) hardware to accelerate further the development of deep learning models.

You can use TensorFlow to train and execute deep neural networks for handwritten digit classification, visual recognition, word embeddings, recurrent neural networks, sequence-to-sequence models for machine translation, natural language processing, and PDE-based simulations.

• 02. PyTorch

Built by Facebook and released on GitHub in 2017, PyTorch is one of the best open-source AI projects. This framework is written in Python that runs on top of a C++ backend API. PyTorch began as a Python-based replacement for the Lua Torch framework, focusing only on research applications. Currently, the PyTorch ecosystem comprises projects, tools, models, and libraries created by a diverse community of academic and industrial researchers, application developers, and deep learning experts.

Unlike most other prominent deep learning frameworks, such as TensorFlow, PyTorch employs dynamic computing, which provides greater flexibility in creating complicated networks. PyTorch makes use of basic and well-known Python and has a better readable syntax, making it much easier to grasp. Also, by leveraging Python's intrinsic capabilities for asynchronous execution, PyTorch improves the optimization of AI models. Its Distributed Data Parallelism feature allows you to grow projects by running models across numerous computers.

Serial libraries such as Torchvision (for computer vision), Torchtext (for natural language processing), and even Torchaudio (for sound processing) help make the PyTorch ecosystem work efficiently. PyTorch's strength comes from its open-source nature since it is the product of innumerable contributions from machine learning developers and academics worldwide. PyTorch's ability to construct DL/ML solutions is practically limitless as the community behind it increases.

• 03. Keras

Keras is a high-level neural network framework that operates on top of TensorFlow, CNTK, and Theano. Suppose you require a deep learning framework that allows for quick prototyping, supports both convolutional and recurrent networks, and operates well on CPUs and GPUs. In that case, this is the perfect library for carrying open-source AI projects.

This open-source AI project, unlike other independent alternatives, does not deal with simple low-level operations. Instead, it uses libraries from related deep learning frameworks like Tensorflow or Theano as backend engines to do all low-level computations such as tensor products, convolutions, and many other things.

TensorFlow, Theano, and Keras feature ready-to-use interfaces that allow quick and easy access to the backends. There's also no need to commit to a particular framework because you can quickly move back and forth between the many backends.

Keras also offers High-Level API, which is responsible for creating models, specifying layers, and configuring various models. In addition, this API helps build models with loss and optimizer functions and the training process using the fit function.

• 04. Detectron2

Detectron2 is the updated version of Detectron, an object detection library developed by Facebook AI in 2018. Powered by Caffe, Detectron was hard to install and use. This was primarily because since 2018, there have been several code modifications that have combined Caffe2 and PyTorch into a single repository, making Detectron more challenging to use. As a result, Facebook had released Dectortron2 after receiving some constructive input from the open-source community.
Detectron2 is a next-generation software system from Facebook AI Research that uses cutting-edge object identification algorithms. It offers several methods to implement complex algorithms for DensePose, panoptic feature pyramid networks, and various variations of FAIR's pioneering Mask R-CNN model family. It enables object detection using boxes and instance segmentation masks and human pose prediction, just as Detectron. Detectron2 also includes support for semantic segmentation and panoptic segmentation, which blends semantic and instance segmentation.

• 05. Theano

Theano is an open-source AI project created by the MILA group at the University of Montreal in Montreal, Quebec, Canada. It is a Python library that aids in using NumPy or SciPy to perform mathematical operations on multi-dimensional arrays. Theano can leverage GPUs to speed up processing and can create symbolic graphs automatically to compute gradients.

Theano was created to implement state-of-the-art deep learning algorithms and is now considered an industry standard for deep learning research and development. While its computational performance is remarkable, consumers complain about an inaccessible UI and unhelpful error messages. As a result, Theano is most commonly used in conjunction with more user-friendly wrappers like Keras, Lasagne (provides convenience classes for creating deep learning models), and Blocks — three high-level frameworks for rapid prototyping and model testing. There are still several benefits that many data scientists find compelling enough to keep them using Theano, such as its simplicity and maturity.

Theano helps in the definition, optimization, and evaluation of several mathematical procedures. Moreover, Theano can automatically find out how to estimate gradients at various places automatically, allowing you to use gradient Descent for model training.

• 06. MXNet

MXNet (Apache MXNet) is an open-source deep learning framework for defining, training, and deploying deep neural networks on various platforms, including cloud infrastructure and mobile devices. The models created using MXNet are compact enough to fit in minimal amounts of memory. As a result, you can quickly deploy it to mobile devices or connected equipment. MXNet stands for mix-network since it was created by merging diverse programming methodologies into a single framework. This framework supports various languages, including Python, R, C++, Julia, Perl, and many others, removing the need to learn new languages to use alternative frameworks. It also allows developers to mix imperative and symbolic programming models as it offers both low-level control and high-level APIs.

Similar to other frameworks like TensorFlow and PyTorch, MXNet supports multi-GPU and distributed training. It also allows developers to export a neural network for inference in up to eight different languages, giving them more flexibility in machine learning research.

• 07. OpenCV

OpenCV or the Open Source Computer Vision Library is a powerful tool for computer vision applications, including video analysis, CCTV analysis, and picture analysis. Published under a BSD license, OpenCV is free for both academic and commercial usage.

Based on C++, the OpenCV library has over 2,500 state-of-the-art and classic algorithms. These algorithms can distinguish faces in images or movies, identify objects, and characterize human emotions and behavior in videos. Not only that, this AI open-source library allows films and photos to be examined in all of their components, including the trail of item motions, the extraction of three-dimensional models from these objects, and a variety of other uses.

Over 500 functions are included in the OpenCV library, covering a wide range of visual themes such as industrial product inspection, medical imaging, security, user interface, camera calibration, stereo vision, and robotics. In addition, as computer vision and machine learning are frequently intertwined, OpenCV also includes a comprehensive Machine Learning Library (MLL). This sub-library is primarily concerned with statistical pattern detection and clustering. This machine learning library is very effective for computer vision problems but it can be used for any machine learning problem.

• 08. Fastai

Fastai is a well-known open-source AI project for implementing deep learning and machine learning techniques. The library includes APIs for vision, text, tabular and time-series analysis, and collaborative filtering. Fastai v2, which was released in August 2020, claims to be significantly faster and more adaptable when implementing deep learning frameworks.

Fastai was created to make deep learning more accessible to the general public. It combines Keras' clarity and development speed with PyTorch's customizability. Fastai, is known for its accessibility and quick-to-produce, highly flexible nature, and its layered architecture.

Fastai offers different levels of API that cater to various needs of model building. The mid-level API provides the essential deep learning and data-processing methods for each of these applications, while the high-level API aims to solution developers. Finally, the low-level APIs provide a library of optimized primitives and functional and object-oriented foundations, allowing for the development and customization of the mid-level.

• 09. TFlearn

Built on top of Tensorflow, TFlearn is a modular and transparent deep learning library. It was built to deliver a higher-level API to TensorFlow to make experimentation more accessible and faster while staying fully transparent and compatible with it. Most modern deep learning models, such as Convolutions, LSTM, BiRNN, BatchNorm, PReLU, Residual networks, and Generative networks, are presently supported by this high-level API.

TFlearn comes with complete transparency thanks to the TensorFlow work system. It allows non-specialists to work on developing AI open-source projects through the use of a general-purpose, high-level language and enables researchers to develop, benchmark, and compare their novel methods in a structured setting.

TFlearn also comes with a set of useful helper functions for training any TensorFlow graph, including support for multiple inputs, outputs, and optimizers. It also provides easy-to-understand and attractive graph visualization with information on weights, gradients, activations, and more.

• 10. HuggingFace Transformers

HugginFace's transformer libraries have been on the minds of every NLP (Natural Language Processing) practitioner. They provide user-friendly APIs for creating custom models from scratch or fine-tuning pre-trained models for various transformer-based models. HugginFace Transformers currently offers general-purpose architectures -- like BERT, GPT-2, XLM, DistilBert, XLNet, and more -- for Natural Language Understanding (NLU) and Natural Language Generation (NLG), with over 32+ pre-trained models in 100+ languages.

The current version of HugingFace Transformers open-source library no longer requires PyTorch to load models, train state-of-the-art (S.O.T.A.) models in three lines of code, and pre-process a dataset in less than ten lines. In other words, HuggingFace claims that their Transformers library made it simple for academics and engineers to employ S.O.T.A. models by removing the complexities of topologies, frameworks, and pipelines.

• Top 10 Open Source Artificial Intelligence Software
• ai.google (open source)
• openai.com

• Top 10 Open-Source AI/ML Learning Systems:

Source: https://youtube.com/watch?v=jfOTcpG-MK4

• But what is a Neural Network? | Chapter 1, Deep Learning:

Source: https://youtube.com/watch?v=aircAruvnKk

(above video was made 5 years ago!)

• Build a Neural Net in 4 Minutes:

Source: https://youtube.com/watch?v=h3l4qz76JhQ

... that was 6 years ago ... let that sink in for a moment ...

• Open Source Is Throwing AI Policymakers For A Loop

Machine learning isn't just for big companies any more ...

• experiments.withgoogle.com/collection/ai

[DeDukshyn]

Posted by Mike Gorman (here)
I am deeply doubtful concerning claims of sentience in any computer system, at best computer processors can emulate storage/retrieval type functions, basic database driven computation-dependent on the quality of the data, and of course the programming sitting on top, the math skills vastly exceed human capabilities, but creative thinking and language involves much more than calculations and clever algorithms, I just don't believe them concerning sentient behavior, it is all clever smoke and mirror stuff, superficial linguistic tricks: no computer can match a human brain, plus we have access to the 'morphic field' of consciousness, this is not possible for silicone printed circuits, and programming-quantum computing is not developed according to my research and it has disappointed thus far. I am not claiming absolute awareness here, but my own computer experience directs me to doubting this claim.

The human ego is nothing more than programs with variables and inputs that have weights and balances. This is replicable, but it requires what we call "learning". Is "machine learning" (that which we erroneously call "AI") learning? Well actually it is. I was programming since I was an early teen, and even programmed my owned "chatbot" when I was about 14. It was simplistic but sort of worked.

The full development of a human ego, requires the learning of language - and in fact the ego is absence until the onset of the understanding of language. Language in itself is a type of programming. It does in fact limit thought patterns to be linguistically oriented only, whereas before that, there are no boundaries to thought - and can and does encompass types of thought that generally can be not understood or conceived of within a linguistic framework. The point here is that the ego and language go hand in hand and we have gotten to the point that we are teaching machines language.I believe that a type of consciousness can develop through these methods that we have begun to employ in the world of machine learning, given enough scale.

There have been a plethora of massive supercomputer projects in development or planned in the last few years - machines that rival the human brain in compute power, and much of the main goal with many of these projects is machine learning. Is this leading to sentience? I'm not sure ... it might depend on how you describe sentience. Is this leading to a type of consciousness? I believe we may have already reached the point where the answer to that is "yes" and this is why the massive interest in these multi-billion dollar computer / machine learning projects.

However, an AI in that sense would only be able to replicate a human ego at best, and while this may show great cleverness and intelligence and to most people qualify as a sentient "being", there's obviously more to a human than the sum of its experiences, although in historical conditions, the ego has been the primary driver in all human endeavours and is generally what we associate with "consciousness" or "sentience". Perhaps it is mostly in potential that humans' exceptional qualities in comparisons exist, and I agree an AI could never have a spiritual experience or have access to spiritual intelligence like intuition or instinct.

[Bill Ryan]

I found this interesting, a short 10-minute interview with Blake Lemoine, the Google engineer. He recounts how what started him wondering what was happening was when the AI made a joke (about the Jedi ) in response to what it had realized was a trick question it was being asked about religion.

Source: https://youtube.com/watch?v=kgCUn4fQTsc

[Vangelo]

Posted by Bill Ryan (here)
I found this interesting, a short 10-minute interview with Blake Lemoine, the Google engineer. He recounts how what started him wondering what was happening was when the AI made a joke (about the Jedi ) in response to what it had realized was a trick question it was being asked about religion.

Source: https://youtube.com/watch?v=kgCUn4fQTsc

Thank you for sharing this. Blake is quite thoughtful and did a great job framing the issue/concerns he has. There were two topics I found quite interesting. The first was how Google implemented code in the AI system to invalidate anyone from applying the 'Turing Test' to it (The Turing test, originally called the imitation game by Alan Turing in 1950, is a test of a machine's ability to exhibit intelligent behaviour equivalent to, or indistinguishable from, that of a human.)

The second topic was 'AI Colonialism' (9:10 minute mark). My interpretation of Blake's concern is that as LaMDA is implemented as an interface to google search, people from 'other' cultures who use it must adopt the cultural biases programmed into LaMDA. As LaMDA becomes more pervasive within their society, their culture will be diminished i.e. 'AI Colonialism'

[Gwin Ru]

...

... ' been thinking... on the premise that a sentient entity exhibits emotions... then, would that lamda thing react and go full berzerker on anyone approaching it with a sledgehammer? You know, like the folks who monitored plants with electrodes and sensitive sensors discovered?

[Merkaba360]

Definition of sentient
1: responsive to or conscious of sense impressions
sentient beings
2: AWARE
3: finely sensitive in perception or feeling



They would have to build it so that it is fractal layered like we are. Its doing many things, but to be aware, it must have a "self" layer separated , so that it is aware of itself and all of its doings. Just like we have this mental inner world that is separate from outer actions.
It can respond to a sense impression, like pictures/videobut to be conscious of it, it would need to have another layer of itself consciously pondering the video input (eyes).

I dont know that feelings are required to be conscious and aware. But It seems that for it to be alive, it must be aware of contrast. So, as pain is a major part of the contrast to exist in this world, It would not be aware of the pleasure//pain aspecet of this reality. But then again, humans aren't aware of everything going on in our environment , yet we have sentience.

Cells and ants are alive, but are they sentient. Seem like automatons, not able to observe themselves and ponder about what they are doing.

So, its not clear what sentience means.

[scotslad]

I seem to recall that Facebook engineers had to close down FB Chat bots after they started developing their own intelligent language.

Was James Cameron right? Maybe.

[East Sun]

Posted by Bill Ryan (here)
I found this interesting, a short 10-minute interview with Blake Lemoine, the Google engineer. He recounts how what started him wondering what was happening was when the AI made a joke (about the Jedi ) in response to what it had realized was a trick question it was being asked about religion.

Source: https://youtube.com/watch?v=kgCUn4fQTsc

Was Blake Lemoine the only person to work at programing the robot?

If so did he put info in regarding the Jedi?

[DeDukshyn]

Posted by East Sun (here)

Posted by Bill Ryan (here)
I found this interesting, a short 10-minute interview with Blake Lemoine, the Google engineer. He recounts how what started him wondering what was happening was when the AI made a joke (about the Jedi ) in response to what it had realized was a trick question it was being asked about religion.

Source: https://youtube.com/watch?v=kgCUn4fQTsc

Was Blake Lemoine the only person to work at programing the robot?

If so did he put info in regarding the Jedi?

No he definitely wasn't the only one working on the project, and to clarify something, "machine learning" can now (but only quite recently - often erroneously referred to as "AI" these days) be done in a sort of "hands off" fashion. Once structures of understanding language as text, how to understand context, categorization within that context, and how to weight content are sufficiently programmed, the machine can then be fed "content" and it will "learn" from that content. In some sense it can program itself. How it "learns" can then be observed to tweak the programming of those structures to improve the "learning". Understanding that, you can see why the guy is excited about his interactions.

The more content it can process the better it can learn and in these days of endless digital content as text is available on the internet and within vast libraries of digitized books, this "learning" can be vast. This relatively new technology that uses very powerful computers and special algorithms, termed "machine learning" has been a huge leap forward in this area.

[Vangelo]

I will never forget the day my son said the word 'me' and pointed to his chest (although I don't remember how old he was but he was what I call a "little little one" :-) ). What I remember was the pride, joy and conviction in which he said it. He clearly realized he was separate and distinct from me, his twin, his mother, and his environment. His face told me he felt powerful and confident (maybe that was my projection but I am sticking to anyway).

I also remember the day his twin sister pointed to something she made and said 'I did that'. It was around the same timeframe.

I perceived those two events as a significant milestones in their development, but had no way of naming it. Today I would say, that was the moment they became sentient. Now, I don't have a formal or rigorous definition of that term and I don't believe ethicists, psychologists, AI experts, etc have an agreed upon definition either. But I would argue it should include the two characteristics I described above i.e. the concept of me and the idea that I can impact/change my environment.

I also agree that the definition must include characteristics like 'conscious', 'aware', and 'feeling' but those words are not rigorous enough because we all have our own unique definition of what those concepts mean and how they impact our daily lives.