"Hal's Legacy" is a seminal book and a companion documentary film about the HAL 9000 computer (Hal) from Stanley Kubrick's 1968 film 2001: A Space Odyssey. Created by Arthur C. Clarke and Kubrick, Hal is an iconic artificial intelligence (AI) in science fiction - a sentient, conversational machine capable of speech, vision, emotion, and even malice.
Released in the late 1990s, these works critically examine how close real-world technology had come to realizing such a system. They blend nostalgia for the groundbreaking movie with rigorous analysis of computing's progress in the 1990s, highlighting both achievements and shortcomings in fields like AI, robotics, and human-computer interaction. Of course, now we are in 2025: "Hal's Legacy" was produced more than a quarter century ago, and the masterwork of Kubrick and Clarke was produced more than half a century ago, and AI technology is in a new accelerating phase.
The book: HAL's Legacy: 2001's Computer as Dream and Reality
Edited by David G. Stork, a computer scientist and expert in pattern recognition, the book "Hal's Legacy: 2001's Computer as Dream and Reality" was published in 1997 by MIT Press. It is a collection of essays contributed by leading figures in computer science, AI, philosophy, and film, inspired by Hal's fictional capabilities. The volume reflects on the relationship between science fiction's visions and technological reality, using Hal as a benchmark to assess late-20th-century advancements. Stork's introduction sets the stage by noting how 2001 treated science with respect, making earnest predictions about the future, though it faltered in anticipating AI's trajectory.
The essays are organized around Hal's key attributes. For instance, chapters explore speech recognition and synthesis, where experts like Ray Kurzweil discuss how computers in the 1990s could generate human-like voices but struggled with natural conversation. Vision and image processing are dissected by Azriel Rosenfeld, who contrasts Hal's ability to read lips and interpret visuals with real systems' limitations in noisy environments. Marvin Minsky, a pioneer in AI, contributes on machine intelligence, arguing that Hal's "mind" embodies early dreams of general AI but overlooks the field's shift toward specialized systems.
Other sections tackle Hal's more provocative traits, such as emotions and ethics. Philosopher Daniel Dennett examines whether machines could possess consciousness or feelings, using Hal's breakdown as a metaphor for AI reliability. The book also covers practical aspects like reliability in space missions, with NASA engineers comparing Hal's fault-tolerant design to actual spacecraft computers. Chess-playing ability is highlighted, referencing Deep Blue's 1997 victory over Garry Kasparov as a real-world parallel to Hal's game with astronaut Frank Poole. Broader themes include the cultural impact of 2001, with film critics analyzing Kubrick's visual storytelling and Clarke's narrative foresight.
Throughout, the tone is optimistic yet cautious: while computing power had exploded per Moore's Law, true Hal-like integration of senses, reasoning, and autonomy remained elusive in the 1990s (and still remain elusive today. The book concludes that 2001's predictions were remarkably accurate in hardware but overly ambitious in software, urging continued innovation. The book is accessible to both experts and enthusiasts, blending technical depth with engaging anecdotes.
The film: 2001: HAL's Legacy
Complementing the book is the 2001 documentary film 2001: Hal's Legacy, directed by David G. Stork and produced as a TV movie. Running approximately 56 minutes, it poses the central question: "How close are we to building a real HAL-9000 computer?" The film weaves archival clips from "2001: A Space Odyssey" with footage from research labs and interviews from the 1990s, creating a dynamic visual narrative that mirrors the book's essays but in a more cinematic format.
At the time of writing, the full film is available on YouTube.
Stork structures the documentary around Hal's functionalities, much like the book. It opens with Arthur C. Clarke reflecting on his creation, admitting that while space travel advanced as predicted, AI lagged due to unforeseen complexities in human-like cognition. Interviews feature luminaries such as Marvin Minsky, who discusses AI's foundational challenges, and Ray Kurzweil, demonstrating speech tech prototypes.
Visual elements bring the analysis to life: scenes from 2001 are juxtaposed with real demos, such as robots navigating environments or voice assistants responding to commands. The film visits labs at MIT, NASA, and IBM, showcasing advancements in computer vision, natural language processing, and robotics. It addresses Hal's darker side, including ethical dilemmas in AI decision-making, with experts debating scenarios akin to Hal's infamous "I'm sorry, Dave, I'm afraid I can't do that."
Unlike the book's textual depth, the film uses narration and graphics to explain concepts like neural networks or lip-reading algorithms. It highlights cultural legacy, with film historians praising Kubrick's design influence on modern interfaces. The documentary concluded on a forward-looking note, suggesting that while a full Hal was distant, elements like virtual assistants were steps toward it.
Interconnections and enduring Impact
The book and film are intertwined; Stork adapted many essay contributors into interviewees, making the documentary a visual extension of the printed work. Both celebrated 2001's prophetic accuracy while critiquing overestimations in AI sentience. They underscored how fiction inspires reality: Minsky credits the movie for sparking his AI career.
In retrospect, "Hal's Legacy" captures a transitional moment before the internet era's AI boom.
Today, with advancements in machine learning and large language models (LLMs), Hal seems less fantastical. Together, the book and film offer a thoughtful tribute, blending entertainment with education, and remain relevant for understanding AI's evolution from dream to partial reality.
The view from 2025
Of course, the book and the film are dated. And how could it be otherwise?
Today, AIs beat all human players at games like chess and the much more complex go. Speech recognition and synthesis, computer vision and image recognition, are essentially solved problems. Generative AI technology like LLMs and image/video generators has been exploding exponentially in the last few years and doesn't seem to be slowing down.
However, the book and film are still valuable when it comes to reflecting on the history of AI technology, its present, and its future.
In an interview titled titled "Computers, Science, and Extraterrestrials," which is included in the book but not in the film, Stephen Wolfram suggests that AI challenges, including pattern recognition tasks like scene analysis, should be addressed using simple, emergent systems like cellular automata rather than complex engineering. He suggests that intelligence emerges from lower-level computations.
In hindsight this seems, or at least it does to me, a description of the purely connectionist approach to AI technology exemplified by today's LLMs: take a very simple neural network, train it with very big data, and you'll be enthusiastically surprised by the results.
"All those cognitive things are just icing on the cake," said Wolfram, "not fundamental at all."
More recently, in his 2023 book "What Is ChatGPT Doing … and Why Does It Work?" - one of the first and in my opinion still one of the best explanations of today's LLM technology - Wolfram concludes that it is "a great example of the fundamental scientific fact that large numbers of simple computational elements can do remarkable and unexpected things" and can shed light on "that central feature of the human condition that is human language and the processes of thinking behind it."