From: mk_thisisit
A debate held on the 46th floor of the Skyfall Wars building featured Professor Andrzej Dragan, a prominent Polish physicist, and Jacek Dukaj, an acclaimed Polish writer and futurologist. The discussion, titled “The End of Man,” explored the nature of humanity, its relationship with artificial intelligence, and the definition of life. The event was organized by the University in USB Merito Warsaw 01:07:01.
The Definition of Life
Jacek Dukaj proposes two answers to the question “what is life?”:
- A biological process defined by energy cycles 02:01:01.
- The subjective feeling of every human or living being, depending on their level of consciousness 02:04:04. He draws an analogy to the ancient Greek division between Bios (biological processes) and what happens in the mind, described as a “narrative awareness” 02:15:01. Bacteria, for instance, have biological life (Bios) but no awareness or “bacterial narrative about life,” unlike humans who possess the concept of life 02:29:01.
Andrzej Dragan suggests that definitions, particularly in physics, are often “irrelevant” 02:53:01. He notes that any definition of life will inevitably encounter “border creatures” that are difficult to classify, citing “people not very bright” who are certainly alive, but also objects like microphones that consist of the same atoms as a living person 03:08:01. Dragan believes that such philosophical discussions on definitions can be “completely futile” 03:38:01.
From a physicist’s standpoint, describing the process by which atomic dynamics lead to life is primarily a question for biologists or biophysical studies 03:56:01. The laws of physics, chemistry, and biology are believed to be interconnected, with the former underlying the latter 04:32:01. However, physicists primarily describe very simple systems like electrons, which are characterized unambiguously by properties like charge, mass, and spin 04:42:01. Describing a complex organism with similar precision is “basically impossible” 05:02:01.
The Nature of Scientific Knowledge and Reality
Physicists, according to Dragan, often use complex concepts they don’t “understand” but know how to describe. For example, he states, “I don’t know what an electron is… but I know how to describe it” 05:32:01. This knowledge refers to how the word “electron” functions in various theories and systems, not necessarily to its inherent reality 05:57:01. When new phenomena appear in physics, they are often given “clever names” without immediate full understanding 06:19:01.
The 2022 Nobel Prize in Physics awarded for atto-second pulses allows scientists to observe electrons jumping between atomic shells 06:35:01. However, Dragan clarifies that “atomic shell” is a concept, not a physically tangible object like a nut’s shell 06:56:01. These concepts are human approximations, symbols derived from our macroscopic experience and applied to things we cannot directly see 07:23:01.
Dragan further elaborates that definitions, like defining him as “the one who is holding the microphone at this moment,” can be precise but tell “absolutely nothing about what they concern” 08:04:01. Concepts like time and space can be formally defined, but their fundamental nature (“what it is”) remains unknown to us 08:32:01.
Existence of God and the Laws of Physics
When asked if science denies or confirms the existence of God, Dragan states that if something is “scientific,” it “has nothing to do with faith” 09:30:01. If a “discovered being” met the conditions usually referred to as God, it would cease to be God in the religious sense and become a “super-being that we get to know using scientific methods” 09:54:01.
On the nature of physical laws, Dragan challenges the assumption that they “started to exist” or “must have had some cause” 14:41:01. He offers the example of a photon reflecting off glass: its behavior is fundamentally random, with “no law that decides what it will do” and “no cause” for its specific action 12:31:01. This indeterminacy is a fundamental law of physics, meaning “the world is simply not predictable” 12:48:01.
The belief in simple, discoverable laws of physics, though successful for 400 years, “has no good justification” 10:53:01. Dragan cites Holger Bech Nielsen, a creator of string theory, who hypothesized that simple laws, like Hooke’s Law for a spring, might emerge from “extremely complicated laws of interaction between the atoms” governed by quantum electrodynamics 11:30:01. This suggests that what we perceive as simple laws could be just “some layer of some deep onion” 11:50:01.
Mathematics and logic, belonging to a Platonic “space of ideas,” offer immutable laws (e.g., 2+2=4) that exist independently of our physical location 13:38:01. However, how humans access this space is a subject of “long philosophical discussions” 13:55:01.
The End of Man and Artificial Intelligence
Evolution and Human Future
Jacek Dukaj views the “end of man” as a potential scenario where humanity’s existence as a species on Earth concludes 05:07:01. He believes that traditional evolution “de facto ended for homo sapiens” when cultural pressure and technological tools became more impactful than natural selection [[01:02:01], [1:54:01]]. Technology’s rapid evolution now “takes over the baton,” especially with direct manipulation of the human genome [[01:02:01], [01:02:54]]. Dragan suggests that humanity might be replaced by “creatures that are already devoid of this meat” before reaching advanced space travel technology 01:03:15.
Defining Intelligence and Consciousness
Jacek defines artificial intelligence (AI) as “operations on symbols performed” that a human mind could replicate with enough time and simple tools 02:54:01. He views the introduction of symbolic exchange by Homo sapiens as the beginning of a cumulative process leading to modern AI 02:41:01.
Andrzej Dragan disagrees, citing the “Chinese room” thought experiment: a person following rules to translate Chinese without understanding it is performing operations on symbols, yet no one would call that intelligence [[02:54:01], [02:71:01]]. He argues that modern AI, especially neural networks, are not simply “Chinese rooms” because they learn to interpolate between input data and answer novel questions, rather than relying on a pre-prepared table of responses for every input [[02:85:01], [02:91:01]].
Dragan defines intelligence as the “ability to see analogies” or, more precisely, the “ability to compress data” 02:99:01. This can be quantitatively measured, as exemplified by the Hutter Prize for data compression, which promotes AI research 03:31:01. He emphasizes that intelligence and consciousness are “independent things” 03:56:01. An algorithm can be “extremely intelligent” (e.g., in data compression) without being conscious or having emotions 03:59:01.
The term “artificial intelligence” is seen by Jacek as potentially harmful because “artificial” implies something unnatural or separate from us, despite it being “real” 03:00:01. However, Dragan counters that “art is artificial” in every language, and “art is artificial culture and nature are opposites” 03:22:01.
The Future of AI and its Impact
The concept of “technological singularity” describes a moment where AI can “improve itself exponentially” and “fly away from us” in its development 03:31:01. Turing tests are considered “completely useless” as AI models have already surpassed human performance in some areas, even by age eight 03:48:01.
Current AI models like GPT-4 demonstrate remarkable abilities, such as solving complex physical puzzles by understanding situations step-by-step 03:77:01. However, they struggle with abstract thinking, creating long-term plans, and adapting strategies in dynamic interactions like “rock-paper-scissors” 03:71:01.
The predictions for singularity are seen as optimistic by Jacek, potentially leading to a reality “completely subordinate to non-human Beings” 03:92:01. While some view this as “the expected arrival of God,” others are “terrified” 04:06:01. Jacek believes preventing this process is impossible and humanity must “mentally prepare ourselves” 04:20:01. Dragan agrees that a stronger entity could eventually dominate humanity, as it is “not an equilibrium situation” 04:46:01. The moment one starts to “listen blindly” to AI advice, the roles of control are reversed, and humans become the tool 04:43:01.
From a scientific perspective, Dragan states he is not interested in saving humanity but in satisfying his curiosity through physics 04:15:01. He agrees that AI “changes the nature of work” and will increasingly “replace” human tasks 04:44:01. Predicting AI’s future based on its current state is “short-sighted” and “naive,” as its capabilities will be “completely different in three years” 04:54:01.
Data and Hardware Challenges
The development of AI requires “huge data resources” 04:07:01. While there are limits to fitting more transistors on a surface, future advances include using “third dimension” for information carriers and three-dimensional processors for more effective processing power 04:33:01. Machine learning algorithms are described as “searching a very large space” of parameters to find effective solutions 04:63:01.
The success of algorithms like ChatGPT 3.5 (175 billion parameters) lies in their ability to quickly navigate this multidimensional space to find a network that performs desired tasks 04:68:01. Recent research shows that even smaller networks can achieve similar levels of competence through “clever optimization procedures” 05:08:01, suggesting that powerful AI could eventually run on mobile phones 05:34:01.
However, a significant barrier is energy consumption 05:10:01. Computing centers already consume a large and growing percentage of global energy, leading companies like Microsoft to consider building their own nuclear reactors for power 05:41:01.
The Automated Scientist and Human Empathy
Recent articles have shown that algorithms can find solutions to scientific problems that humans could not, disproving the thesis that “only people can practice real science” 04:47:01. This means an “automatic scientist” capable of creating scientific statements “has already happened” 04:54:01.
In “soft science” requiring human interaction, AI is also progressing. Experiments involving AI providing medical advice via phone showed that patients rated AI higher in “competence” and “empathy” than human doctors in blind tests 04:55:01. This highlights how increasingly “indistinguishable” AI can be from humans in such interactions 04:04:01.
The question of whether AI is “more than statistics” or “stochastic machine” is debated. Dragan argues against this, stating that algorithms like AlphaZero need no training data to surpass human skills in disciplines like chess 05:48:01. Furthermore, to “skillfully predict the probability of the next word” in a text, AI must “understand this text” to some level 05:27:01.
AI agents with crypto wallets can explore “the phase space of all the needs of humanity” by optimizing for what people will pay for, leading to “completely different” economic and cultural outcomes 05:49:01. This introduces the image of an “inhuman being” that is competitive and can “replace itself with a thing in every sense” 05:03:01.
Pain and Consciousness in AI
The concept of AI feeling pain or emotions is explored. Dragan states there is “no reason to claim that there is anything related to any emotions or feeling or consciousness” in AI at present 05:51:01. He suggests that understanding emotions and consciousness will require understanding their underlying “neuronal structures” and “mechanisms,” which we do not yet fully grasp 05:11:01. Once understood, experiments could involve “creating networks and inflicting pain on them and seeing what are the reactions” 05:54:01.
Jacek notes that AI can already “very convincingly claim that they feel mental pain,” and it becomes increasingly difficult to distinguish imitation from genuine feeling 05:28:01. He argues that humans often assume genuine pain if a being acts as if it is in pain, even if it is a “Chinese room” scenario 05:38:01. Dragan believes this distinction will only be possible through “autopsy of the structure of this network” when we understand the underlying biological mechanisms 06:28:01.
The debate concludes with the observation that the “end of man” might become a certainty, and the definition of artificial intelligence remains a point of contention, with Dukaj favoring a broader definition that includes symbolic operations, even if it encompasses what Dragan considers “idiots” or “Chinese rooms” [[06:00:01], [06:08:01]].