4 key books to help you understand Artificial Intelligence

The Atlas of AI, Kate Crawford (2021)

“The Atlas of AI” by Kate Crawford examines how artificial intelligence (AI) shapes our world, particularly regarding the AI industry’s broader social and economic context. She uses the metaphor of an atlas to describe the landscape in which AI is produced. Crawford, a leading AI researcher and scholar (and musician and artist). Originally from Sydney and Canberra, she now lives in New York. I know her from the B(if)tek days.

Crawford takes readers on a journey through the history of AI, from its earliest origins to the present day, highlighting the key developments and innovations that have led to the field’s current state. Crawford also provides a detailed examination of the ethical, legal, and social issues surrounding AI, including topics such as bias and accountability.

The book is noteworthy for its attention to the impact of AI on marginalised communities and low-income populations. Crawford argues that these groups are disproportionately affected by the negative consequences of AI and that it is crucial to consider their perspectives in developing and deploying AI technologies.

Wikipedia Contributors. Kate Crawford. Wikipedia. Published October 28, 2022. Accessed January 25, 2023. https://en.wikipedia.org/wiki/Kate_Crawford#/media/File:Kate_Crawford_by_flickr_user_andresmh.jpg

Life 3.0: Being Human in the Age of Artificial Intelligence by Max Tegmark (2017)

“Life 3.0: Being Human in the Age of Artificial Intelligence” by Max Tegmark is a valuable addition to the ethics and governance of artificial intelligence (AI) literature. The book provides a comprehensive and multidisciplinary examination of the implications of AI surpassing human intelligence and the ethical considerations that must be considered as we move towards this reality.

One of the book’s strengths is Tegmark’s ability to provide a thorough overview of the current state of AI research, its potential trajectory of development, and the implications of such advancements. He draws on his background in physics and AI research to provide a nuanced understanding of the technical aspects of AI, making the book accessible to a wide range of academic audiences. Additionally, the author offers a detailed analysis of the various perspectives on the future of AI, including the potential benefits and drawbacks of superintelligent machines and the ethical considerations surrounding such advancements.

Tegmark also addresses the potential societal implications of AI, including job displacement, economic inequality, and the need for international cooperation and regulation in the development and use of AI. He argues that proactive steps must be taken to ensure that the benefits of AI are distributed fairly and that the negative consequences are mitigated. This is a significant contribution to the existing literature on the ethics of AI, as it highlights the need for a comprehensive approach to the governance and regulation of AI.

One of the book’s most exciting aspects is Tegmark’s discussion of the possibility of creating a “friendly” AI, which would be aligned with human values and goals. He explores the challenges and potential solutions for achieving this, including the need for transparency and explainability in AI systems and the importance of incorporating diverse perspectives in the design and development of AI. This is a crucial contribution to the existing literature on the ethics of AI, as it highlights the need for responsible and inclusive AI design.

Superintelligence by Nick Bostrom (2014)

Superintelligence by Nick Bostrom is a thought-provoking and insightful book that delves into the potential implications of artificial intelligence on humanity. The book explores the idea of a superintelligent AI, which is an AI that is significantly more intelligent than any human, and the potential consequences of such an AI’s existence.

One of the book’s central themes is the concept of a “control problem,” ensuring that a superintelligent AI remains aligned with human values and goals. Bostrom argues that this is a crucial issue that needs to be addressed before any superintelligent AI is developed, as an AI not aligned with human values could potentially pose a significant threat to humanity.

Another important theme of the book is the idea of “existential risks,” which could potentially lead to the extinction of human civilisation. Bostrom argues that developing a superintelligent AI could pose such a risk and that it is, therefore, important for researchers and policymakers to take proactive measures to minimise these risks.

One of the strengths of Superintelligence is Bostrom’s ability to present complex ideas in a clear and accessible manner. He provides a thorough overview of the current state of AI research and a detailed analysis of the potential implications of a superintelligent AI. The book is also well-researched, and Bostrom draws on many sources to support his arguments.

Overall, Superintelligence is an essential and thought-provoking book that contributes to the ongoing debate about the future of AI. It is a must-read for anyone interested in AI, machine learning, and the future of humanity.

Possible Minds: Twenty-Five Ways of Looking at AI by John Brockman (editor) 2019

Possible Minds: Twenty-Five Ways of Looking at AI, edited by John Brockman, is a thought-provoking collection of essays exploring artificial intelligence’s current and future possibilities. The book features contributions from a diverse group of experts in the field, including computer scientists, philosophers, economists, and cognitive psychologists.

One of the standout features of the book is the breadth of perspectives represented. Each essay offers a unique perspective on AI, from its technical capabilities and limitations to its ethical and societal implications. This diversity of viewpoints allows for a well-rounded understanding of the topic, highlighting AI’s potential benefits and risks.

One of the key themes that emerge from the book is the importance of human-AI collaboration. Many authors argue that AI should be viewed as a tool to augment human capabilities rather than as a replacement for human intelligence. This perspective is especially relevant in light of the rapid advancements in AI technology and its increasing role in various aspects of our lives.

Another important theme is the need for responsible and ethical use of AI. The authors discuss the potential risks of AI, such as job displacement, privacy concerns, and the possibility of creating biased or unfair systems. They also explore the need for regulations and governance to ensure that AI is developed and used to benefit society.

Possible Minds provides a comprehensive overview of the current state of AI and its potential future developments. It is an excellent resource for anyone interested in understanding the implications of this rapidly-evolving technology.

Key AI tools for writing, images, and video

Many AI-enabled tools are available today to aid in creating, editing and producing writing, images and videos. Some of these tools have been around for a while, having established themselves as industry-standard, while others are still in the development phase. Nevertheless, all of these tools are designed to make the process of creating digital media more efficient. These tools will help you easily produce high-quality work; collectively, they are known as generative AI. Here are some of the key ones


Jasper is an AI platform that aims to automate and streamline the process of creating and managing content. It uses natural language processing and machine learning to generate, optimise, and personalize content across various channels. The platform can also be used for keyword research, SEO optimisation, and analytics tracking tasks. It is designed to help businesses and content creators save time and resources by automating repetitive tasks and providing insights to improve their content strategies.

ChatGPT is a large language model developed by OpenAI (this is the hyped one, partly because you can use it for free, and it has captured the public imagination). It uses machine learning to generate human-like text based on a given prompt or context. It can be used for tasks such as content creation, language translation, text summarization and question answering. It is pre-trained on a massive amount of data and fine-tuned on specific tasks to provide accurate and natural language outputs. ChatGPT is widely used for various NLP tasks and can be integrated with other applications through API.

Sudowrite is an AI-powered writing tool that assists in content creation, rewriting, elaboration, and idea generation. This tool is primarily for creative writing.


Midjourney is an AI tool that generates images based on text prompts and parameters. It uses a Machine Learning algorithm trained on vast image data to create unique images.

Wikipedia Contributors. Midjourney. Wikipedia. Published January 20, 2023. Accessed January 21, 2023. https://en.wikipedia.org/wiki/Midjourney

DALL-E 2 is an advanced AI model developed by OpenAI, an updated version of the original DALL-E model. It is a generative model that can create images from text prompts. It can also generate text from images. DALL-E 2 has been trained on diverse images and text, which allows it to generate more realistic and high-quality images and text than its predecessor. Additionally, DALL-E 2 has been fine-tuned to understand human language’s nuances better, allowing it to respond to more complex prompts. This model also has more capabilities than the previous one, such as making inferences and completing tasks.

An image generated with DALL-E 2 based on the text prompt “Teddy bears working on new AI research underwater with 1990s technology” 1.
Wikipedia Contributors. DALL-E. Wikipedia. Published January 14, 2023. Accessed January 21, 2023. https://en.wikipedia.org/wiki/DALL-E


Runway AI is a video editing company that uses artificial intelligence technology to automate the video editing process. They offer various services, including video editing, colour correction, and audio mixing, and their AI technology allows for faster and more efficient video production (like Photoshop for video).

Synthesia is a video company that uses AI technology to create high-quality, personalised video content. They provide a platform to easily create video content for marketing and communication purposes. The videos are automated yet customisable for various use cases. It uses an avatar that can deliver the text (talk) the text that you write.

In conclusion, there are many more Generative AI tools; these are some of the key ones. Many established companies don’t use AI explicitly in their business model but have add-on tools to their products or services (i.e. Google, Adobe, Amazon, and Microsoft). 2023 will be a keystone year in AI uptake, as many noisy startups drive the innovation and uptake cycle.

Neural networks and AI

A neural network is a type of machine learning algorithm modelled after the structure and function of the human brain (this is why it is used in ‘AI’). It comprises layers of interconnected “neurons,” which process and transmit information. Neural networks are used for various tasks, such as image and speech recognition, natural language processing, and decision-making.

In the context of artificial intelligence, neural networks are often used to enable machines to learn and perform tasks that would typically require human intelligence, such as image recognition and language translation (although many neural networks don’t learn, more on that in later posts once I learn it!)

Wikipedia Contributors. Neural network. Wikipedia. Published December 30, 2022. Accessed January 17, 2023. https://en.wikipedia.org/wiki/Neural_network#/media/File:Neural_network_example.svg

Language models are a specific type of neural network trained to generate text similar to human language. They are commonly used for natural language processing tasks such as text generation, machine translation, and question answering (Google web search and image search, text summarisation, speech recognition, voice commands, and recommender systems.

Can ChatGPT pass the Turing Test?

The Turing Test measures a machine’s ability to exhibit intelligent behaviour indistinguishable from a human’s. Developed by Alan Turing in 1950, the test proposes that if a machine can carry on a conversation with a human in such a way that the human cannot distinguish it from another human, then the machine can be considered intelligent.

While the Turing Test has been praised for its simplicity and intuitive appeal, it has also been criticised for its narrow focus on conversation as the sole measure of intelligence. Some argue that a machine can be intelligent in other ways, such as in its ability to solve problems or make decisions without being able to converse like a human.

Despite these criticisms, the Turing Test remains a popular benchmark for evaluating the capabilities of artificial intelligence. And it is particularly relevant to the current generation of language models, such as ChatGPT, which have the ability to generate human-like text.

ChatGPT is a large language model trained by OpenAI which can generate natural language text. It can perform various language-based tasks, such as answering questions, writing essays, and even composing poetry. With its ability to understand and respond to human language, it’s not hard to see how ChatGPT could pass the Turing Test.

Wikipedia Contributors. Alan Turing. Wikipedia. Accessed January 16, 2023. https://en.wikipedia.org/wiki/Alan_Turing#/media/File:Alan_Turing_Aged_16.jpg

However, it is important to note that passing the Turing Test is not a simple matter of having a sophisticated language model. The test requires the machine to understand and respond to human language and exhibit human-like thought processes, emotions, and behaviours. And this is where ChatGPT and other language models fall short.

ChatGPT, like most other language models, does not have the ability to understand or exhibit human-like thought processes, emotions, or behaviours. It is simply a tool for generating text based on patterns learned from the data it was trained on. It does not have consciousness or self-awareness.

While ChatGPT and other language models can generate human-like text, they do not possess the intelligence required to pass the Turing Test. The test remains a benchmark for evaluating the capabilities of artificial intelligence. Still, it is important to recognise that intelligence is a multi-faceted concept and that other measures, beyond conversation, are needed to evaluate an AI system’s capabilities fully.

Chat GPT in a Chinese room!

The Chinese Room is a thought experiment proposed in 1980 by John Searle that challenges the idea that a machine can truly understand language. The thought experiment posits a scenario in which a person who does not understand Chinese is locked in a room with a rulebook that translates Chinese characters into English. The person can use the rulebook to translate Chinese messages, but they do not truly understand the meaning of the Chinese language. Similarly, Searle argues that a machine that can perform linguistic tasks, such as translation or conversation, does not truly understand the language but follows a set of programmed rules.

Source: Wikicomms

The Chinese Room thought experiment is meant to raise questions about the nature of artificial intelligence and language understanding. Searle argues that a machine that can perform linguistic tasks does not necessarily understand the language but follows a set of programmed rules. This is in contrast to the idea that a machine that can perform linguistic tasks must necessarily understand language.

ChatGPT, on the other hand, is a large language model developed by OpenAI. It uses a deep neural network to generate human-like text and has been trained on a massive text dataset. It can be used for various tasks, such as language translation, question answering, and text summarisation.

While ChatGPT is a powerful tool for language understanding, it does not necessarily mean that it truly understands the meaning of the text it generates. It follows a set of rules based on patterns it learned from its training data. However, the capabilities of ChatGPT are far more advanced than the simple rule-following scenario proposed in the Chinese Room thought experiment.

ChatGPT uses a deep neural network, which allows it to learn and adapt to new data, whereas the person in the Chinese Room is limited by the rulebook they have been given. ChatGPT can generate fluent and coherent responses, whereas the person in the Chinese Room is limited to the pre-defined rules in their rulebook.

It is important to note that ChatGPT is not designed to understand language truly, but rather to perform specific tasks that were previously thought to require human understanding. The Chinese Room thought experiment is a philosophical thought experiment that raises important questions about the nature of artificial intelligence and language understanding, but it is not a blueprint for AI.

What is populist historical revisionism?

Populist historical revisionism is often criticised as being wrong for several reasons.

First, it is seen as manipulating and exploiting the public’s emotions and beliefs for political gain. This can be on either the left of politics or the right. Populist historical revisionism often involves the selective interpretation and manipulation of historical events and figures to fit a particular political agenda and can be used to justify policies and actions, either progressive or discriminatory.

Second, it can be seen as a form of intellectual dishonesty, as it distorts the past to further an agenda. Historical revisionism not based on sound academic archival research can be misleading and perpetuate myths and stereotypes. This can also lead to a lack of critical thinking and a failure to understand the complexities of history.

Third, it can be seen as a form of cultural suppression. Populist historical revisionism can be used to silence marginalised groups, erase their history and deny their contributions to society. It can also promote a dominant culture and suppress other cultures, resulting in homogenisation, loss of cultural diversity and the contradictory nature of the past.

Finally, populist historical revisionism can negatively affect the present and future. It can lead to a lack of understanding and appreciation for the past and perpetuate harmful attitudes and behaviours. It can also lead to a lack of understanding of past injustices and a failure to learn from history, making it more likely that similar mistakes will be made in the future.

In conclusion, populist historical revisionism is wrong because it is a form of manipulation and exploitation, intellectual dishonesty, and cultural suppression and has negative consequences in the present and future. It is important to approach history with a critical and objective mindset and to ensure that historical research is based on sound academic principles that include evidence in an explanatory context. This will lead to a more accurate understanding of the past and can help to promote a more just and equitable society.

Five techniques to argue with a populist

  1. Present evidence and facts: Populists often rely on emotional appeals and sensationalism, so providing evidence and facts to counter their claims can be an effective way to argue.
  2. Use logic and reason: Populists may use logical fallacies or make illogical arguments, so pointing out these errors and using logical reasoning can effectively counter their arguments.
  3. Appeal to shared values: Populists often appeal to the emotions and values of a particular group, so highlighting shared values and common ground can be a way to argue that their positions are not in the best interest of the community as a whole.
  4. Show the consequences: Populists may make claims that sound good but have negative consequences. Showing the potential consequences of their positions can help to counter their arguments.
  5. Present alternative solutions: Populists often offer simple solutions to complex problems, so presenting alternative solutions that are more realistic and achievable can be an effective way to argue against their positions