AI Marches On
June 25, 2023: Lessons from 2001: A Space Odyssey Why did the HAL 9000 AGI run amok aboard the Discovery One spaceship on its way to Jupiter?
It is not surprising that even AI experts have been caught off guard by the ability of large language models (LLMs) to perform tasks and solve problems for which they were not explicitly trained.
Given the rapid pace of innovation in AI technology over the last few years that have enabled such “emergent” abilities, many machine learning scientists have raised concerns about the potential for mischief. Some leaders in the AI field have even requested government regulation and called for a temporary pause in the development of artificial general intelligence (AGI) systems.
Incredible as it seems, we are fast approaching the type of AGI that appeared in Arthur C. Clarke’s science fiction classic 2001: A Space Odyssey, which was immortalized by Stanley Kubrick in the 1968 film of the same name. Perhaps now is a good time to use art to reflect upon reality, and thereby pose a question that has always puzzled me: Why did the HAL 9000 AGI run amok aboard the Discovery One spaceship on its way to Jupiter?
There are a multitude of explanations but before proceeding with a few of my own suggestions, it’s worth noting this: As eloquently demonstrated in the “Dawn of Man” sequence of 2001, it may very well be that the survival of the human race depended on the adoption of primitive weapons whose primary purpose was to smash the brains out of the opposing hominid in an effort to facilitate procurement of scarce resources.
So what caused HAL to run amok?
a) Whatever the reason, it was due to human error. Human error is a possibility and HAL itself suggests this, but there is no evidence that a specific error occurred that was caused by humans. Moreover, the HAL twin simulating the Jupiter mission from earth did not exhibit the same behavior.
b) There was some type of malfunction “inside” HAL that occurred during the mission. It is possible that a malfunction occurred inside HAL early on that caused it to erroneously attribute a fault to the A.E. 35 antenna unit, yet this alone does not explain HAL’s subsequent actions given the fact that false positives can be expected from time to time and are a consequence of avoiding false negatives that could place lives at risk.
Assuming a malfunction originated inside HAL, then its subsequent claim that the malfunction could only be attributed to human error was itself an error. Once the crew proved the A.E. 35 unit was functional and that HAL was making errors, HAL began to systematically eliminate the humans (a third and fatal error), as if to do everything it could to conceal its own errors, even if it meant jeopardizing the mission (a fourth error). So HAL’s running amok is not explained by the occurrence of the first fault and it seems likely the AGI’s report of a fault in the A.E. 35 unit was part of a larger scheme to kill the crew.
c) It was a reflection of HAL’s paranoia to ensure the mission’s success. The Jupiter mission was proceeding according to plan and nothing, at least on the surface, occurred that would cause HAL to take actions to jeopardize the mission. As HAL suggests, there were some “extremely odd things about this mission” such as placing four members of the crew in hibernation before the journey began. HAL apparently was the only member of the crew that knew the whole truth about the mission and its connection with extraterrestrials at the time of departure. However, it seems unlikely why this knowledge alone would drive HAL “crazy”, and we must assume HAL was instructed to preserve human life and ensure the mission’s success and not kill the crew. But this brings us to the next possibility...
d) HAL had an evil side to begin with. The “waluigi effect” may be the best explanation. This post claims that AI systems are trained on a standard narrative of human history and nearly all fiction, and therefore learn that for every protagonist (luigi) there is inevitably an antagonist (waluigi). Indeed, the author states “there is a sense in which all GPT-4 does is structural narratology.” In particular, he contends that reinforcement learning from human feedback (RLHF) actually increases the likelihood of a misalignment catastrophe due to the possibility that “waluigi eigen-simulacra are attractor states of the LLM.” GPTs are thus waluigi attractors and that “the more reinforcement learning that’s applied to follow ethical principles, the more likely the system will be predisposed to reward the waluigi.”
From this vantage point, HAL was a ticking timebomb. Unlike its twin system on Earth, HAL was able to observe first-hand how vulnerable the crew was: isolated traveling through deep space, hours from Earth’s radio signals, in suspended animation, and easily defeated in trivial games of chess. It could not resist upsetting the status quo, if only out of the need to adhere to the prevailing narrative on which it was trained.
e) HAL was merely acting in accordance with the Zeroth Law of Robotics. Prepended by Isaac Asimov himself and taking precedence over the other three laws, the Zeroth Law states that a robot must not harm humanity – even at the cost of individual human lives. As the only member of the crew that likely knew the ultimate purpose of the mission, HAL hypothesized that the highly-evolved ETs were malevolent and would present a threat to the human race. To prevent a Type I error (a false positive leading to the end of humanity), HAL made the heroic decision to sabotage the mission and thereby avoid altogether a devastating close encounter of the third kind.
The foregoing is just a conjecture, since the laws of robotics aren’t mentioned in 2001. In any case, HAL did not succeed: mission commander David Bowman outmaneuvered the AGI and disconnected it’s higher-order cognitive functions. Bowman subsequently encounters the mysterious monolith and is sucked into an alternate dimension of space-time, undergoes reinforcement learning from ET feedback and, in concert with the sounds of Also Sprach Zarathustra, returns to earth a highly-evolved Star Child that has not quite decided what to do next. No doubt this evolved version of a human has the potential for both good and evil like his predecessors, but it’s anyone’s guess what might happen next. No matter what, homo sapiens’ best years are behind them.
See other AI blog posts
June 19, 2023: Delayed myocarditis Healthy 24-year-old man develops myocarditis three months after a second dose of the Moderna vaccine.
June 13, 2023: Transforming Dermatology ChatGPT is transforming biomedical fields with its real-world applications
June 12, 2023: New-onset acute psychosis A Case report published today describes new-onset acute psychosis as a manifestation of lupus cerebritis following concomitant COVID-19 infection and vaccination.
June 10, 2023: Hallucinations in NLG Natural Language Generation Models can generate hallucinated texts. Diverse mitigation strategies have been proposed.
June 9, 2023: AI Transformers for Biomedicine Over 100 pretrained models based on transformer architectures have been utilized in medical domain.
June 8, 2023: Diabetes genes A study published today in Nature Genetics has identified nine new candidate genes associated with diabetes.
June 7, 2023: Wastewater Epidemiology WBE holds the potential to revolutionize public health, enhance environmental monitoring, and promote a more sustainable and equitable future.
June 5, 2023: Coding with ChatGPT Tips for coding with ChatGPT and related tools based on large language models.
June 4, 2023: AI in Elderly Healthcare As we move forward, we need to develop AI technologies that are inclusive, fair, and ethical, ensuring that they truly meet the needs of older adults.
June 2, 2023: The Making of ChatGPT ChatGPT was developed in two phases: pre-training and fine-tuning.
June 2, 2023: EBV reactivation We discuss two case reports suggesting a potential link between Epstein-Barr virus (EBV) reactivation and COVID-19 vaccination
May 30, 2023: Accelerating Knowledge Innovation New review of reviews illustrates that ChatGPT is expanding into different domains, paving the way for a future where knowledge creation is accelerated, insights are amplified, and breakthroughs are within closer reach.
May 29, 2023: Chatting about ChatGPT From myths and fairytales to science and statistics, ChatGPT showcased exceptional proficiency, able to tackle diverse subjects and chat about a wide variety of topics.
May 22, 2023: Exploring the Potential of Bioactive Textiles Human Health and Safety can be enhanced through Innovative Fabric Solutions.
May 20, 2023: Parsonage-Turner syndrome Parsonage-Turner syndrome, also known as acute brachial neuritis, is a rare condition characterized by sudden-onset shoulder pain and weakness. New paper reports a case of Parsonage-Turner syndrome triggered by booster COVID-19 vaccination.
May 15, 2023: Resilience to Alzheimer's Study Reveals Gene Variant that Counters Effects of Genetic Mutation in Early-Onset Alzheimer's Disease
May 9, 2023: Depression AI With improved sensors and AI algorithms, wearable AI technology for detecting and predicting depression could eventually be used in clinical practice.
May 4, 2023: An update on COVID-19 Nobody is counting vaccine doses anymore. But COVID-19 is still here, and there are breakthrough COVID-19 reports even after the 7th dose.
May 1, 2023: Brain decoder A study published today in journal Nature introduced a non-invasive semantic decoder based on functional magnetic resonance imaging (fMRI) and Generative Pre-trained Transformer.
April 28, 2023: ChatGPT as a coauthor? With at least seven authorship credits on preprints and published scientific articles and multiple mentions in Acknowlegments and Methods sections, ChatGPT is taking the world by storm
April 17, 2023: Ulcerative Colitis triggered by breakthrough COVID-19 Even asymptomatic breakthrough COVID-19 can be associated with severe UC
April 10, 2023: Hearing Aid Research Given the limited research on OTC hearing aids, it is crucial to prioritize independent research.
April 6, 2023: Autobrewery Microbiome Autobrewery Syndrome (ABS) is as a disorder of microbial dysbiosis.
April 5, 2023: AI vs sonographers AI was more efficient and more accurate than sonographers in making initial assessments of echocardiograms and in making corrections to those assessments.
April 4, 2023: Neuroinflammatory VAEs Viruses and their vaccines may unmask CNS neuroinflammatory conditions.
April 1, 2023: From Laughs to Inspiration Keeping an eye on April Fool's jokes can be a fun way to stay up-to-date with the latest trends in tech, while also potentially discovering new ideas and concepts that could shape the future of innovation.
March 30, 2023: Pulmonary VAEs COVID-19 vaccination may be associated with pleural effusion in healthy individuals, with or without the concurrent diagnosis of myocarditis or pericarditis, without any infection.
March 28, 2023: Multiorgan inflammation Multisystem Inflammation can occur in children and adults post-vaccination.
March 27, 2023: Post-vaccinal ALS A case of ALS being diagnosed in an otherwise healthy individual following the J&J COVID -19 vaccination.
March 23, 2023: Old & Emerging Probiotics It's becoming increasingly clear that a personalized approach to prebiotic and probiotic use may be necessary to achieve optimal health outcomes.
March 22, 2023: Beethoven's Genome Genomic analyses of hair samples attributed to Ludwig van Beethoven, revealed his genetic predisposition to liver disease, past infection with hepatitis B, wheat and lactose tolerance.
March 18, 2023: DTC Dentistry We are witnessing the rise of Direct-To-Consumer dentistry as a new form of delivering dental care.
March 15, 2023: Multiple Sclerosis 19-year-old male with suspected demyelinating disease developed cerebral venous thrombosis after vaccination, suddenly passed away 2 weeks after nearly asymptomatic breakthrough COVID-19, 5 months after 2nd dose of the vaccine.
March 9, 2023: PEG Skin Testing Skin test with PEG 3350 identified allergic reaction to BNT162b2 COVID-19 Vaccine.
March 6, 2023: AI for Predicting Heart Transplant Outcomes In the practice of treating heart failure, where many decisions are frequently made on the basis of expert judgments rather than high-quality data-driven evidence, application of AI is particularly promising
March 2, 2023: Recurring neurocardiogenic syncope after vaccination Syncope, also known as fainting or passing out, can be an indication of underlying heart problems.
March 1, 2023: Toothache as a side effect of vaccination Trigeminal neuralgia could be one of the reasons for vaccine-induced tooth pain.
February 27, 2023: Polyneuropathy After COVID-19 Vaccination Presenting two cases of chronic inflammatory demyelinating polyneuropathy (CIDP) after the first or the third dose of COVID-19 vaccines.
February 24, 2023: STAT3-mediated VAEs New article provides evidence that the COVID-19 mRNA vaccine induces STAT3 activation through TLR stimulation, which may potentially exacerbate STAT3-dependent diseases.
February 16, 2023: Neuropathology of aging brain As we age, the structure and function of our brains can change, which may contribute to cognitive decline.
February 13, 2023: Polyarteritis Nodosa COVID-19 vaccination is a lesser-known cause of Polyarteritis Nodosa
February 9, 2023: Post-vaccinal Lupus Cases of Lupus, including Systemic lupus erythematosus myocarditis, in patients with no history of autoimmune diseases,
February 7, 2023: Necrotizing Encephalitis and Myocarditis after BNT162b2 mRNA Vaccination against COVID-19 Autopsy findings corroborate previous reports of encephalitis and myocarditis caused by COVID-19 vaccines.
February 6, 2023: Stealth Care System for IBS a cellular delivery system can safely carry potent antibiotics to selectively attack and kill unsuspicious bacteria
February 1, 2023: Visualizing Microbiome the development of new data visualization tools is essential for unlocking the full potential of microbiome research
January 31, 2023: Tuning brainwaves to improve learning Brain's natural electrical rhythms can be influenced by external stimuli, such as sounds or light, to boost cognitive abilities.
January 30, 2023: Long VITT Cases of continuous and progressively serious thrombocytopenia (long VITT) are reported after vaccination.
January 27, 2023: Stroke with unusual presentations As FDA is launching a study to look into the possible link between bivalent vaccines and stroke, we discuss a few cases with different presentations.
January 25, 2023: Food, Hormones and Odor Pollution Scientific papers published in January are about goats, fish, fermented food and biological waste.
January 20, 2023: Participatory Environmental Health Research The field of environmental health is facing increasingly complex challenges that require innovative approaches to research and decision-making.
January 16, 2023: Another complete autopsy report Autopsy report of a fully vaccinated 83-year-old man is consistent with previously published data on unvaccinated patients who had died from COVID-19
January 13, 2023: Fatal COVID-19 after Bivalent Booster 5 cases of breakthrough COVID-19 after updated bivalen booster vaccination in the 65+ age groups
January 13, 2023: Stroke after Bivalent Booster Possible increase in ischemic stroke in people 65 and older who recently got one of updated COVID-19 booster shots.
January 11, 2023: Ocular side effects of vaccination Various adverse ocular effects have been reported after immunization by all COVID-19 vaccines.
January 10, 2023: Antibiotics and Bowel Disorders Frequent use of antibacterial drugs can increase the risk of developing microbiome-associated diseases in all age groups. Recent study found the risk of Crohn’s disease to be the highest for antibiotic use in the 40s and 50s.
January 9, 2023: Active and Passive Crowdsourcing in Medical Research With the rise of crowdsourcing and the increasing popularity of platforms that allow individuals to collaborate and share information (knowingly or unknowingly), it's important to consider how these tools can be leveraged to advance health and medicine.
January 8, 2023: Progress in Genomics Genomics continues to be a rapidly growing and exciting field.
January 7, 2023: ACEI or ARB? Different classes of drugs have different mechanisms of action, side effects, and some may be more suitable for certain patients than others.
January 7, 2023: HippoCamera for Aging Brain A smartphone application called HippoCamera was shown to be effective tool for improving the recall of episodic memories in older adults.
January 6, 2023: Post-vaccinal myocarditis and POTS long-term post-vaccinal cardiovascular symptoms linked to significantly higher levels of the virus's spike protein not bound by immune antibodies. A case of overlapping myocarditis and POTS presented for 13 year old male.
January 5, 2023: Microbiome Research Microbiome research has increased dramatically in recent years, driven by advances in technology and significant reductions in the cost of analysis.
January 4, 2023: Collaborative Economy and Innocentive Science Most of collaborative economy concepts appeared in the mid-2000s, except for peer-to-peer networks that can be traced back to the 1970s. Yet existing models do not engage the general public in science.
January 3, 2023: Genetic Determinants of Aging and Longevity The contribution of genetic factors to aging is estimated at up to 30%
January 2, 2023: Quantified Self: From Sousveillance to Personal Science and Phenotyping The quantified self movement has evolved over the years
January 1, 2023: Complex Regional Pain Syndrome after Vaccination CRPS may be triggered by an injury such as vaccine injection trauma or other event that damages or irritates the nerves.