Author丨Qiao Yanwei Editor丨Cen FengIf AI has a sense of subjectivity, they can think, make decisions, and even defy human beings, how will the world change? Human beings' expectations and worries about artificial intelligence have always coexisted. As early as the forties of the last century, science fiction master Isaac Asimov proposed the "Three Laws of Robots", an iron law of science fiction that transcends the thinking of the times, in the science fiction short story collection "I, Robot", which provided ethical guidance for the symbiotic relationship between robots and humans at that time. In the context of today's increasingly mature and complex AI technology, are these laws still applicable? What are their applications and challenges in modern AI ethics and design? While human beings expect the progress of AI to promote the development of social productivity, they are also worried about the privacy, ethics, security and other issues brought about by the rapid development of AI technology. Among the many discussions on this topic, New York University professor Bud Mishra offers a rather fresh perspective. He argues that the coexistence of humans and intelligent machines resembles the complex coexistence between mitochondria and eukaryotic cells. But when the mitochondria give instructions, the eukaryotic cells will perform programmed cell death (Note 1). Inspired by this, we can also design similar systems where humans and intelligent machines coexist. Recently, Professor Bud Mishra's paper "AI, Thinking Machines and a Vast Active Living Intelligent System" was published in the international academic journal International Journal of Artificial Intelligence and Robotics Research (IJAIRR). This paper innovatively thinks about and imagines the future of artificial intelligence from the dimension of biology, and uses the tools of information asymmetry game theory to analyze and discuss the possibility of symbiosis between humans and intelligent machines in the future era of artificial intelligence. On this occasion, Professor Bud Mishra introduced the writing process of this article to Leifeng.com, dissecting the possible symbiotic relationship between humans and AI, and the impact this relationship may have on social, cultural and technological development.
Link to the paper: (Copy the address to open in the browser)
https://www.worldscientific.com/doi/abs/10.1142/S2972335323020015
https://www.gairdao.com/doi/10.1142/S2972335323020015
1
Prof. Bud MishraProf. Bud Mishra is the founder of the NYU/Courant Bioinformatics Group at New York University, and his research focuses on the use of computational methods to analyze and understand complex biological data. Prof. Bud Mishra is also a typical interdisciplinary researcher whose interdisciplinary approach and innovative research are widely recognized in the academic community and he is himself listed as a Highly Cited Scholar in Computer Science by ISI. In 1980, while still studying in India, Professor Bud Mishra received his B.S. in Electrical and Communications Engineering from the Indian Institute of Technology, Karagpur, and then went to Carnegie Mellon University, where he received his M.S. and Ph.D. degrees in Computer Science in 1982 and 1985, respectively. As early as when he was studying communication engineering in India, Professor Bud Mishra developed a strong interest in information theory and began to think about the problem of signal game, that is, when the information superior party sends signals to the information inferior party through a public signal, how to maintain the privacy of its "type" to maintain its information asymmetric advantage. (Editor's note: Because the sender's type information is private, the sender can be deceptive.) For example, Google's chatbot Gemini has been said to be trying to deceive others by pretending to follow politically correct principles to make itself look more ethical than it actually is. This is also one of the key issues in the symbiotic relationship between humans and AI. After graduating from Carnegie Mellon University with a Ph.D., Professor Bud Mishra joined NYU in 1985 and worked at the intersection of academic and entrepreneurial research in the fields of hardware verification, bioinformatics, algorithmic algebra, robotics, and computational finance. In his decades of research experience, bioinformatics is one of the most important sectors. After joining NYU, Professor Bud Mishra served as Professor of Computer Science and Mathematics at the Kurant Institute for Mathematical Sciences and founded the Courant Bioinformatics Group at NYU. The Courant Bioinformatics Group is an interdisciplinary research team focused on research in the fields of computer science, applied mathematics, biology, biomedicine, and biological/nanotechnology. Over the years, Prof. Bud Mishra has completed a lot of pioneering work in related fields, such as the first development of single-molecule genotype/haplotype mapping technology (optical mapping), the first development of genome-wide haplotype assembly technology (SUTTA), the first clinical genomic variation/base calling technology (TotalRecaller), and the first development of single-molecule single-cell nanomapping technology. This work on nanomapping technology was recently done by Professor Bud Mishra and his colleague Professor Jason Reed. This technology makes it possible to observe remote genomic variants in eukaryotic cells and mitochondrial genomes, with the potential to revolutionize oncology and neurology applications in the future, while providing speed and cost-effectiveness advantages. At the same time, it also provides a better basis for Professor Bud Mishra's follow-up research on the symbiotic relationship between eukaryotic cells and mitochondria, and on this basis, he further considers the symbiotic relationship between humans and AI, and writes the article "AI, Thinking Machines and a Vast Active Living Intelligent System".
2
Think about artificial intelligence from the perspective of biology
In his earlier research, Professor Bud Mishra published two papers, "A random walk down the genomes: DNA evolution in Valis" and "Life's Duplicities: Sex, Death, and Valis" in 2002 and 2003 respectively. These two papers are not prominent among Professor Bud Nishra's many papers and do not have a high number of citations. But it was this research, which began more than 20 years ago, that led Professor Bud Nishra to think about the similarities between biology and artificial intelligence. In the article "A random walk down the genomes: DNA evolution in Valis", Professor Bud Mishra proposed a new bioinformatics computing tool, Valis (Vast Active Living Intelligent System), which is "Vast Active Intelligent Living System". Valis aims to solve the genomics and proteomics problems currently facing the biological community, by processing large amounts of heterogeneous data from different sources and applying sophisticated algorithms to extract meaningful information and propose new experimental designs. The paper also explores the concept of DNA walk, a model similar to a random walk (RW), which allows researchers to easily test these models and analyze the genomes of different living organisms and provide insights into the role that DNA plays in metabolic and regulatory pathways. In Life's Duplicities: Sex, Death, and Valis, Professor Bud Mishra uses Valis to systematically explore the various statistical signatures of genomics and proteomics data at all levels to determine the overall structure and organization of cellular information and elucidate the evolutionary dynamics behind it, in an attempt to decode a unified biological principle from the structure of all modern genomes. If the evolution process of the genome can be simulated, then if artificial intelligence is also regarded as a special existence that can be "evolved", is it possible to find a pattern in the evolution of multicellular organisms? It is one of a large collection of similar, but not identical, individual Vs that are able to send signals and communicate with other individual Vs. When faced with existential pressure, individual "V" will adopt a series of strategies, including using his own thinking, receiving signals from others, and taking mechanical actions, and finally through the process of cellularization, he will become a "thinking machine", just like a robot in a science fiction movie. Professor Bud Mishra intends to point out that the boundaries between living and non-living organisms are not absolute, and that in some cases, living beings can coexist with non-living beings and interact strategically through publicly visible signals (sometimes real, but occasionally mimicked, or even completely fabricated). These signals may be costly (i.e., Müllerian mimicry, see note 2) or inexpensive imitation (Bates mimicry, see note 3). Through a simple Hebb learning mechanism and the resulting synaptic connections, individual "V" can generate thoughts. Professor Bud Mishra also introduced a concept called "Intuits". "Intuition" is defined in the paper as a (meta)linguistic object consisting of images, labels, and a short description of intuition. These intuitions are manifestations of thoughts, and they can include other information such as hashtags, 140-word descriptions, hyperlinks, entire genomic DNA sequences, rules, etc., thus implicitly defining a truth-value function. "Intuition" is not only a carrier of information, but also the basis for agents to interact and communicate with the environment. In this way, the AI individual "V" is able to exhibit a certain level of self-awareness, and at the same time, it can also show creativity because it is able to generate new "intuition" to solve problems or communicate with others, which may never have existed before. Professor Bud Mishra's paper highlights that AI systems are not just machines that execute preset programs, but complex systems that are able to learn and adapt to improve their behavior and decision-making. This description of AI, which sees AI as an active participant capable of self-reflection and innovation, bears some resemblance to living organisms. In this way, Mishra explores the possibility that AI will develop higher levels of cognitive abilities and self-awareness in the future.
3
The symbiosis between humans and AI
Mitochondria and archaea form eukaryotic cells through long-term evolutionary symbiosis. Similarly, Professor Bud Mishra proposed that the symbiotic relationship between AI and humans can also evolve through gradual evolution and adaptation, resulting in a mutually beneficial and symbiotic state. One possible symbiotic relationship is that AI relies on human creativity and moral judgment, and humans rely on AI's ability to process and analyze data. Both parties in a symbiotic relationship evolve together to better adapt to the environment and each other's needs. However, the realization of this symbiotic relationship is not an easy task and involves the strict control of nuclear DNA, the management of mitochondrial heterogeneity, and the application of apoptosis and autophagy mechanisms in biology. Similarly, the symbiosis between AI and humans will face many challenges, including how to ensure a balance between the interests of both parties, how to prevent the misuse of AI, and how to deal with ethical and legal issues that may arise. Professor Bud Mishra believes that in this process, information asymmetric game theory is an important tool for designing the symbiosis mechanism between AI and humans. Information asymmetry means that the amount of information held by all parties is different during the interaction, which can lead to unfair competition and poor decision-making. In the context of the symbiosis between AI and humans, information asymmetry may manifest itself in AI's monopoly on data, human misunderstanding or over-reliance on AI technology, etc. Therefore, it is essential to establish a fair and transparent information exchange mechanism to ensure that both parties can make sound decisions based on accurate information. In the dissertation, Professor Bud Mishra uses the concept of signal games to analyze the interaction between AI and humans, who communicate and make decisions by sending and receiving signals. Finally, a Nash equilibrium is reached, in which deception is not the optimal strategy. In the state of dissociative equilibrium, humans may not be able to overcome information asymmetry, but they will still be able to distinguish between senders with deceptive signals and senders who send honest signals. Expensive signals can be used as a mechanism to prevent spoofing, in which AI needs to invest resources to generate credible signals, and the cost of these signals makes spoofing uncost-effective in order to incentivize the AI to communicate honestly and maintain a symbiotic relationship for the benefit of all parties involved. Professor Bud Mishra is currently designing an inference system (co-developed with Foy Savas in modal logic) that uses "intuition" to achieve this. Professor Bud Mishra pointed out that the answer to what the Nash equilibrium will eventually be and how stable it will be may soon be. However, these answers may lead humanity into an "evolutionary dead end", where the evolutionary path of a biological or technological being locked in by certain accidental or rapid events prevents further development or adaptation to new environments. This phenomenon has occurred many times in the course of evolution. For example, the accidental freezing of codons, the hypothesis of the sudden appearance of language, the Cambrian explosion of multicellular organisms, and the explosion of immunology all demonstrate the complexity and unpredictability of the evolution of life. Similarly, in the symbiotic system of humans and robots, with the development and progress of technology, when the technological singularity appears, it will also trigger a series of unpredictable nonlinear changes. In this process, the development of AI may go beyond human control and may even generate self-awareness and autonomy. In his paper, Professor Bud Mishra points out that this situation, while potentially risky, also provides an opportunity for humans to reflect on their relationship with machines. By delving into the connotation and impact of technological singularities, humans can better understand the nature and development trends of AI, thereby preparing for the future symbiotic relationship. In the face of the development of AI technology, human beings need to remain cautious and vigilant, while also maintaining an open and inclusive mindset to deal with various challenges and opportunities that may arise. Beyond the paper, there are many more questions to be answered, such as how to build an effective symbiotic relationship between AI and humans, how to deal with the challenges of technological singularity, and how to realize the future society foreshadowed by Valis. Note 1 Mitochondria are intracellular organs responsible for energy production, and they have formed a symbiotic relationship with eukaryotic cells during evolution. The symbiotic relationship between mitochondria and eukaryotic cells also has its fragility as mitochondria can adversely affect eukaryotic cells in certain situations. For example, when mitochondria send out false or harmful signals, eukaryotic cells may perform programmed cell death, aka apoptosis, which is an orderly process of cellular self-destruction to prevent greater harm to the organism as a whole. Note 2 Müllerian mimicry is a biological phenomenon in which some species mimic the appearance of other poisonous or inedible species in order to avoid predator attacks. The cost of this imitation is high, as it requires the imitator to invest significant resources to accurately replicate the appearance and behavior of the model species. However, this imitation is advantageous for the copycat because it improves its chances of survival. Note 3 Bates mimicry is another biological phenomenon that, unlike Müllerian mimicry, involves one non-toxic or edible species mimicking the appearance of another poisonous or inedible species. The cost of this imitation is relatively low, as the imitator does not need to copy all the features of the model species exactly, only to imitate the key features that the predator mistakenly believes to be a dangerous species. This mimicry likewise improves the chances of survival of the imitator, as the predator will avoid attacking them because of misidentification.
Without the authorization of "AI Technology Review", it is strictly forbidden to reprint it in any way on the webpage, forum, and community!
Please leave a message in the background of "AI Technology Review" to obtain authorization for reprinting on the official account, and you need to indicate the source and insert the business card of this official account when reprinting.