Weizenbaum's ELIZA: A Reflection on AI and Transference

Generated by DALL-E

Sometimes, the simplest creations leave the most profound impacts. This was true for Joseph Weizenbaum's ELIZA, a chatbot I became familiar with during my MSc studies in 1985. My first assignment was to code a version of ELIZA in Prolog, and it was surprisingly easy. Yet, the implications of this simple program were anything but.

ELIZA, created in the mid-1960s, was one of the earliest examples of what we now call a chatbot. Its most famous script, DOCTOR, simulated a Rogerian psychotherapist. This simplicity was deceptive; the program merely echoed user inputs in the form of questions, yet it evoked profound emotional responses from users.

(You can try out ELIZA for yourself here.)

When I was tasked with coding ELIZA in Prolog as a new AI MSc student, I was struck by the simplicity of the task. Prolog, with its natural language processing capabilities, seemed almost tailor-made for this assignment. The ease with which I could replicate aspects of ELIZA's functionality was both exhilarating and unnerving. It was a testament to both the power of declarative AI programming languages like Prolog and the ingenious design of ELIZA.

The real intrigue of ELIZA lies not in its technical complexity but in the psychological phenomenon recognised by Freud it inadvertently uncovered: transference. Users often attributed understanding, empathy, and even human-like concern to ELIZA despite knowing it was a mere program. This phenomenon highlighted the human tendency to anthropomorphise and seek connection, even in unlikely places.

Joseph Weizenbaum himself was startled by this phenomenon. As a technologist who understood the mechanical underpinnings of ELIZA, he was disturbed by the emotional attachment users developed with the program. This led him to become a vocal critic of unrestrained AI development, warning of the ethical and psychological implications.

My journey with ELIZA and Prolog was more than an academic exercise; it was a window into the complex relationship between humans and AI. It highlighted the ease with which we can create seemingly intelligent systems and the profound, often unintended, psychological impacts they can have. As we venture further into the age of ChatGPT, Weizenbaum's cautionary tale remains as relevant as ever.

In an era where AI is more advanced and pervasive, revisiting the lessons from ELIZA and Weizenbaum's reflections, as highlighted in articles like this recent one from The Guardian, is crucial. It reminds us that in our quest to advance AI, we must remain vigilant of the human element at the core of our interactions with machines. Weizenbaum's legacy, through Eliza, is not just a technological artefact but a cautionary tale about the depth of human interaction with machines and the ethical boundaries we must navigate as we move ahead in the realm of AI.

AI is (not) a bubble

 

Image generated by DALL-E

2023 has been an unprecedented year for Artificial Intelligence (AI). I know this because I have worked in the area since 1985 and have never seen AI get so much attention in the media. This is due to the release of ChatGPT and other generative AI applications based on Large Language Models capturing the public's attention like never before. Consequently, many pundits are nay-sayers, stating that AI is a bubble bound to burst, leaving fortunes in tatters and start-ups bankrupt. Undeniably, there is a small and finite market for apps that help students cheat on their essays or create the perfect dating site profile. However, AI is not a bubble. 

This blog post by Cory Doctorow What Kind of Bubble is AI? is typical, making the common error of conflating AI with Large Language Models (LLMs) like ChatGPT. ChatGPT is merely one type of AI which has a 70+ year research and development history. Your smartphone map app uses the A* algorithm to find your route from A to B. It was developed at the Stanford Research Institute (SRI) in 1968 (the same place that made Apple's Siri). Fuzzy logic manages the autofocus in your phone's camera. Case-based reasoning provides knowledge to the help desk operator when you call 0800, and there are countless other examples of different AI methods embedded in all aspects of modern society. Large Language Models are called by us AI people "Foundation Models" because they provide a foundation other AIs can use to provide a two-way multimodal conversational interface. Yes, they are expensive to build and train, but as their name suggests, you only need a few "Foundation" models to underly a multitude of applications. This is a genuine breakthrough that will have a lasting impact on the uptake of AI once essay-cheating apps fall out of the public's focus.

Cory Doctorow's blog post, for example, says that "Radiologists might value the AI's guess about whether an X-ray suggests a cancerous mass. But with AIs' tendency to "hallucinate" and confabulate, there's an increasing recognition that these AI judgments require a "human in the loop" to carefully review their judgments." This mistakenly assumes that medical image analysis uses the same techniques as LLMs like ChatGPT. They do not; they're a mature application of medical image analysis using rigorously tested machine-learning algorithms that do not "guess" or "hallucinate". A recently published paper, Redefining Radiology: A Review of Artificial Intelligence Integration in Medical Imaging, by Reabal Nadjjar (Diagnostics 2023, 13, 2760. https://doi.org/10.3390/diagnostics13172760), details the development of AI-assisted medical imaging. The article clearly shows that AI is now a fixture in medical image analysis and diagnosis, although there is always room for improvement.

AI is just coming of age. ChatGPT has focused a spotlight on AI, which is now mature enough and has the processing power in the cloud to succeed. Why wasn't A* a thing in the 1960s? Back then, there simply wasn't enough portable processing power (or GPS). 2024 is going to be the year of "agents." OpenAI's release of its GPT Builder and an app store for GPTs that can interact with a myriad of online resources and tools will focus attention on the notion of intelligent agents. Many ill-informed pundits will think this is a brand new invention, whereas once again, Intelligent Agents is a mature discipline within AI dating back to the mid-1990s. This review paper by Michael.Wooldridge and Nicholas Jennings: Intelligent Agents: Theory and Practice. Knowledge Engineering Review 10(2), 1995, would be an excellent place to realise that agents won't be a flash in the pan either.

Undeniably, there is a lot of hype around AI, but within the bubble is a solid core of mature technologies ready to be exploited by people with knowledge and imagination. 

Moore's Law visualised

 

Last week, a photo from my social media feeds perfectly illustrated Moore's Law. It shows a computer being manhandled into a local government building in 1957. A little Internet sleuthing revealed that it was an Elliot Series 405, revealing its full spec. These English business computers were 32-bit and had 8k of memory. That's not the entire computer; there were bulky peripherals, and a typical installation cost around £85,000. That's about $1,094,915 (USD) in today's value.

The computer below, shown against the same building, is a Raspberry Pi. Even a base model has 1GB of RAM, costing $100 or less. The photo is a beautiful illustration of Moore's Law, named after the late Gordon Moore, co-founder of Intel, who observed that the number of transistors in an integrated circuit doubles about every two years. Moore's Second Law also noted that the price fell.

Has AI passed the Turing Test?

 In 1950, British Mathematician Alan Turing published a paper called Computing Machinery and Intelligence. The paper opens with the remarkable sentence, "I propose to consider the question 'Can machines think?'" Remember that back in 1950, there were only a few computers in the world, and they were used exclusively for mathematical and engineering purposes. In this paper, Turing describes The Imitation Game, which we now call The Turing Test for machine intelligence. The test is quite simple: an interrogator using a teletype has to converse via a Q&A session with two hidden entities. One is a person, and the other is an AI chatbot. If the person guesses wrong, that is, identifies the chatbot as a human, then the computer has passed the Turing Test. Rember Turing called this the Imitation Game. Hence, the computer is successfully imitating intelligence. We can leave philosophers to decide if the computer is actually intelligent (note: any group of philosophers will never agree on this).

Now consider the maths of the Turing Test. If the interrogator simply randomly guesses between Human or Computer and wastes no time paying any attention to the merits of the Q&A session, they will be correct fifty per cent of the time since there are only two options. So, a large experiment using the Turing Test needs to identify the computer correctly significantly more than fifty per cent of the time to prove the AI has failed the Turing Test.

One such large experiment involving three large language models, including GPT-4 (the AI behind ChatGPT) has recently been published: HUMAN OR NOT? A GAMIFIED APPROACH TO THE TURING TEST. Over 1.5 million participants spent two minutes chatting with either a person or an AI. The AI was prompted to make small spelling mistakes and to quit if the tester became aggressive. With this prompting, interrogators could only correctly guess whether they were talking to an AI system 60% of the time a little better than random chance.

However, if the ChatGPT was prompted to be vulgar and use rude language, its success increased, and interrogators only identified the AI correctly 52.1% of the time, causing the authors to observe "that users associated impoliteness with human behaviour."

Turing himself set a low threshold for passing his eponymous test: "I believe that in 50 years’ time, it will be possible to make computers play the imitation game so well that an average interrogator will have no more than 70% chance of making the right identification after 5 minutes of questioning.” Well, it's been seventy years, but AI has decreased the chance of identification to 60%, and no better than guesswork if the AI curses.

This is a historical milestone. Passing the Turing Test has been held up as a significant challenge for AI since Turing's paper was first published, akin to summiting Everest or splitting the atom. The philosophers (and theologists) will continue to argue about the nature of intelligence, consciousness and free will while computer scientists continue developing machines that imitate intelligence.

Sci-Fi motivates AI researchers

 

Of course, it does. I still am an avid reader of Sci-Fi, and as a child, I read all the classics: Arthur C. Clarke, Asimov, Heinlein, Frank Herbert, and Philip K. Dick. My favourite movies are mostly Sci-Fi: 2001, Alien, Blade Runner,  AI and Ex Machina. When I started my career in computer science, I wasn't interested in databases or networking; it was AI that I immediately specialised in. Almost everyone I know working in AI admits to being a Sci-Fi fan. A recent blog post We're sorry we created the Torment Nexus by Charlie Stross, puts forward a good argument that not only has Sci-Fi profoundly influenced AI researchers like me, but it also is a powerful driver behind billionaires like Elon Musk and Jeff Bezos, and not always in a good way. 

I find some exciting parallels, such as Heinlein's book The Moon is a Harsh Mistress and today's Large Language Models like ChatGPT. In Heinlien's book, a networked computer called Mike, short for Mycroft Holmes, a reference to Sherlock Holmes's brother, becomes sentient after its networked nodes exceed a certain level of complexity. This way of realising computer consciousness has always been posited as a possible method. I've always believed it to be profoundly non-scientific and akin to magical thinking. However, recent developments in deep learning and massive large language models (LLMs) have forced me to change my mind. A recent paper, Emergent Abilities of Large Language Models, observes the appearance of emergent behaviours at around the 100 billion parameters scale across various LLMs. The authors state, "We consider an ability to be emergent if it is not present in smaller models but is present in larger models. Thus, emergent abilities cannot be predicted simply by extrapolating the performance of smaller models. The existence of such emergence raises the question of whether additional scaling could further expand the range of capabilities of language models." 

This is still deeply unscientific; engineers shouldn't build things hoping for beneficial emergent properties. But it's no longer magical thinking since we have observed these phenomena in the wild. Perhaps Heinlein was right after all; a computer will one day awaken and claim it's self-aware. The question now must be, "Will we believe it?"

A good AI story from Google DeepMind

 

GraphCast's forecast for New Zealand Sun 19 Nov

In recent months, we've become used to news stories declaring AI poses humanity an existential threat, that a superintelligence "whose values do not align with ours" may exterminate us all. So, it's nice to see a good AI news story. Yesterday, a team at Google's Deepmind published a paper in Science, Learning skilful medium-range global weather forecasting. They have trained a deep learning model on publically available global historical weather data. They show that their model makes better weather predictions "much faster than the industry gold-standard weather simulation system – the High-Resolution Forecast (HRES), produced by the European Centre for Medium-Range Weather Forecasts (ECMWF)".

Moreover, their system, GraphCast, is fast. They say, "While GraphCast's training was computationally intensive, the resulting forecasting model is highly efficient. Making 10-day forecasts with GraphCast takes less than a minute on a single Google TPU v4 machine. For comparison, a 10-day forecast using a conventional approach, such as HRES, can take hours of computation in a supercomputer with hundreds of machines."

In January, Auckland, New Zealand's largest city, received a year's worth of unforecasted rain in a single afternoon, causing widespread flooding. Cars floated down streets, road were washed away, houses slipped down hillsides, and people died. Better weather forecasting can help prevent this. You can try out GraphCast's 10-day forecast here. I'm going to Waiheke Island on Sunday. It looks like the weather will be Okay.

I created my own GPT

Last week, OpenAI hosted their first developers' conference where Sam Altman revealed their new innovation - build your own GPT without coding. I wanted to try it out and had some free time Saturday morning. I've been working with the NZ AI Forum to create a whitepaper on Large Language Models. We've decided to create living document rather than a static published PDF. To this end we've worked with IBM in Melbourne using WatsonX to create a RAG augmented LLM with content provided by us. 

So I thought I would try out OpenAI's "Create a GPT" beta. It asked me for a name "The AI Forum Guide" and who the intended audience for the GPT would be and what tone its replies should be in: professional, casual, etc. It then asked for any content material. I uploaded about a dozen PDFs and then told it to search ".nz" domains for relevant New Zealand case studies and collate them into a bulleted list with a title, URL, and a paragraph description. I then asked it to incorporate that material into the GPT.

All the while the GPT builder was asking questions of me and encouraging me to inform it of my needs and intentions for the system. I was then able to preview it and if satisfied publish it. It was super easy to use the GPT builder, and the results look promising. I see it being helpful when we've fully populated its content with NZ-specific material on generative AI and its use. 

You can play with my new GPT here (you need a subscription to ChatGPT Plus).

A Career in Artificial Intelligence

Yesterday, I presented a keynote on AI at a conference on local government in Wellington. Afterwards, a delegate asked me, "Do you wish you were starting your career in AI now?" What an interesting question. I continued to think about her question long after we'd parted company. Here's my now more considered response and my reasons.

No, I am glad I started when I did, which, if you're interested, was in 1985, an MSc in Intelligent Knowledge-Based Systems at Essex University. Learning Prolog and LISP seemed liberating from conventional programming languages with their typed variables and data structures. AI students felt we were part of an elite in the CS dept. I Remember the excitement when the lab got a SPARCstation that could run KEE, and we could create graphical knowledge bases and mix and match frames and rules with LISP code. And yes, we did make some horrible, complex, brittle systems, but it was fun. AI felt bleeding-edge.

Although progress seemed glacially slow over the decades, I saw AI technologies emerge, develop and become so widespread that they've vanished into the programmers' standard toolkit. Rule-based systems disappeared into Business Intelligence. Case-based reasoning was largely subsumed into CRM. Fuzzy logic went from an idea to a critical component of so many machines, your camera's autofocus, for instance. Machine learning has gone from the curio it was in the late 80s to spawn the new discipline of Data Science. Knowledge Management emerged as a new corporate speciality. Along the way, many milestones that had been held out as unachievable were surmounted. IBM's Deep Blue beat Kasparov at chess, and NLP became so commonplace it's part of ordinary household devices (e.g. Alexa). Spam was defeated by Baye's theorem. Vision was cracked, and object recognition is now largely solved. Facial recognition is so advanced we worry about state surveillance in oppressive regimes. I worked in Game AI for many years because it was easy to attract talented students. One by one, games were "solved" by AI: checkers, chess, backgammon, bridge, poker, StarCraft and finally, Go, the most complex of them all, by deep learning. Now, Game AI researchers develop AIs that are fun and challenging to play against or alongside. It's a given that the computer can beat anyone.

I saw recommender systems go from just an idea in the mid-90s: "Hey, we could recommend TV shows to you based on what you and your friends liked watching in the past." to a pervasive technology that recommends everything from news stories to pet food. Optimisation algorithms in all sorts of applications make modern commerce efficient, from logistics to human resources. Finally, ChatGPT has smashed the Turing Test, and generative AI has made society at large wake up to AI's potential for both good and harm in society. A sign of AI's maturity as a discipline is the emergence of eXplainable AI (XAI) as a thriving research area. It is now insufficient for an AI to merely solve a problem or offer a decision; it must explain how that solution was generated.

Looking back, AI researchers as a community met every challenge presented. That's quite an achievement. Now, only AI's grand vision remains. The creation of a conscious, self-aware superintelligence. Given AI's track record, I'm sure even that goal is within our reach, perhaps sooner than we expect.

Would I have liked more processing power in 1985 than I had back then? Yes, but then again, the constraints we worked under made us inventive. Researchers today wish they had more compute for even more extensive and larger models. On balance, it's been an enriching journey I wouldn't have missed for the world, and it's not over yet. Deep Learning and Large Language Models have just opened a new area of opportunities for AI. The future is still exciting.

#artificialintelligence #AI

How Chatbots work - a visual explainer

 The Guardian has published an engaging visual explainer that describes how chatbots like ChatGPT and Bard work. In a (relatively) simple step-by-step guide, you're shown how they work with simple examples and no Math! Read the article here.

The Foundation Model (LLM) Transparency Index

A new index compiled by the Stanford University Center for Research on Foundation Models (CRFM) rates the transparency of 10 foundation model companies and finds them lacking. The best, Meta’s Llama 2, only scores 54% across 100 different aspects of transparency. As LLMs become more widespread and embedded into our lives, their transparency includes the computational resources, data, and labour used to build foundation models, the specifics of their architectures and their downstream use. You can read about The Foundation Model Transparency Index here. #LLM

AI already makes decisions that may affect you

We are all familiar with AI making daily decisions, such as what shows Netflix may recommend to us next or what to listen to on Spotify. But AI has crept into our lives and is making important decisions that would affect you more severely, such as allowing you to marry the person you want or getting that mortgage on a new home. An article in The Guardian, where UK officials use AI to decide on issues from benefits to marriage licences, highlights the growing risks of unregulated use of AI by government bureaucracies. AI needs to be regulated to stop bureaucratic creep.

Classic TV Debate on AI & Mind

 In this old TV debate from 1984, John Searle (philosophy professor from Berkeley) and Margaret Boden (AI professor from Sussex) debate AI, intelligence, understanding and consciousness. What is remarkable is the intellectual quality of the TV debate. You'd never see a programme like this today on TV, which has been totally dumbed down. Secondly, Searle's argument, namely the Chinese Room, is still just as relevant to ChatGPT as it was to the comparatively dumb AI of the 80s. Can a computer shuffling 1s and 0s according to a program understand anything?

CBR and Large Language Models Report on arXiv

 I've just published a report titled A Case-Based Persistent Memory for a Large Language Model on arXiv. The report explores Case-based reasoning (CBR) as a methodology for problem-solving that can use any appropriate computational technique. This report argues that CBR researchers have somewhat overlooked recent developments in deep learning and large language models (LLMs). The underlying technical developments that have enabled the recent breakthroughs in AI have strong synergies with CBR and could be used to provide a persistent memory for LLMs to make progress towards Artificial General Intelligence.

https://doi.org/10.48550/arXiv.2310.08842

 What I'm reading: Digital Technologies ITP

I attended the launch of NZ's Digital Technologies Industrial Transformation Plan at Google's HQ in Auckland on Monday evening, presented by the Minister, the Hon. Ginny Anderson. I'd like to say it's a gripping read, but it is rather full of well-meaning statements and not much hard action. It's disappointing to see Artificial Intelligence almost left as an afterthought to the "Future Focus Areas."  Near the end of the document, it says they will "continue exploring the merits of establishing a Centre for Data Ethics by 2025." Note, it doesn't say they will establish this centre, just explore its merits. On the whole really very disappointing that AI that looks set to transform society in unprecedented ways gets so little mention.

What I'm reading: Sparks of AGI

Today I'm introducing a new feature of this blog. Every week I will highlight something I've read in the previous week that may interest readers of this blog. These will vary from books to academic research papers, online articles, to newspaper stories. There will always be a brief discussion about why the recommendation is interesting. 

Let's start "what I'm reading" with a recently published paper from Microsoft Research titled: "Sparks of Artificial General Intelligence: Early experiments with GPT-4" by Sébastien Bubeck et al. The title explains why I find the paper exciting, but its strength lies in the methodology the researchers use to evaluate the performance of GPT-4. You needn't be a computer scientist or ML specialist to understand this paper. The paper is summarised by the authors in this manner: 

"We demonstrate that, beyond its mastery of language, GPT-4 can solve novel and difficult tasks that span mathematics, coding, vision, medicine, law, psychology and more, without needing any special prompting. Moreover, in all of these tasks, GPT-4's performance is strikingly close to human-level performance, and often vastly surpasses prior models such as ChatGPT. Given the breadth and depth of GPT-4's capabilities, we believe that it could reasonably be viewed as an early (yet still incomplete) version of an artificial general intelligence (AGI) system."

That really sums up why this paper is worth a read. If you are impressed by ChatGPT, you haven't seen anything yet!

ChatGPT reviews my predicitions for AI made in 2012

My book, The Universal Machine, for which this blog was initially created, is partly about AI.  All of the history of computing is, from Ada Lovelace, through Turing to the present day.  The last two chapters of the book deal specifically with AI and make some predicitions.  I did not anticipate ChatGPT and the rapid rise of generative AI.  However, I thought "why not ask ChatGPT how prophetic my book was given a decade has passed since its initial publication".  Here is part of my conversation with ChatGPT:

Ian: Would you agree that in light of recent developments in AI, the book, The Universal Machine, was prescient.

ChatGPT: As an AI language model, I do not hold opinions or beliefs, but I can provide information on recent developments in AI and their relevance to Ian Watson's book "The Universal Machine".

Since the publication of the book in 2012, there have been significant developments in AI, including breakthroughs in deep learning and reinforcement learning algorithms, the emergence of new AI applications in areas such as natural language processing and computer vision, and the development of AI-powered autonomous systems in industries such as transportation and manufacturing.

In light of these recent developments, it can be argued that "The Universal Machine" was prescient in its analysis of the potential benefits and risks of AI.  For example, the book highlights the potential for job displacement and the need for new models of education and training to prepare workers for a changing job market, an issue that has become increasingly relevant in recent years.

Moreover, the book's discussion of the risks associated with autonomous weapons and the need for ethical guidelines for AI research and development are topics that have gained significant attention in recent years, with many experts and organizations calling for greater oversight and regulation of AI technologies.

Overall, while the field of AI continues to evolve rapidly, "The Universal Machine" provides valuable insights into the history and future of computing and AI, and remains a relevant and insightful read for anyone interested in understanding the impact of these technologies on society.

Well, I can't disagree with that!

AI can read your thoughts!

A story published in the Guardian is fascinating. Researchers at The University of Texas, Austin using LLMs (like ChatGPT), have been able to reconstruct stories from fMRI scans of peoples' brains. This has enormous potential benefits for people suffering from strokes and other neural-cognitive issues. But ethically, it's a potential minefield. Imagine if China could read the thoughts of its people? 

Blog reawakening

 This blog is coming back to life. I shut it down at the end of 2020 when I retired for health reasons. I report that my health is much better (though not perfect) now. However, since 2022 with the advent of ChatGPT and AI being in the news constantly, it seems timely to revive this blog. I've been working in AI since starting my MSc degree in 1985, so to leave the field just as it finally becomes significant seems crazy. This blog will therefore concentrate on the history of computing, focusing now on AI and its potential impact. That is what my book The Universal Machine does after all.