Smarter Artificial Intelligence: A Not So Obvious Choice

By Shray Ambe

This post was written as part of a class assignment from students who took a neuroethics course with Dr. Rommelfanger in Paris of Summer 2016.

My name is Shray Ambe and I am a rising senior at Emory University. I am a Neuroscience and Behavioral Biology major who is pursuing a career in the medical field. Outside of the classroom, I am involved in organizing the booth for Emory’s Center for The Study of Human Health at the Atlanta Science Festival Expo every year and also enjoy volunteering at the Emory Autism Center and the Radiology Department at Emory University Hospital. 

At the 2016 Neuroethics Network in Paris, France, bioethicist and philosopher John Harris gave a lecture titled “How Smart Do We Want Machines to Be?” During his lecture, Harris discussed the potential impacts of artificial intelligence (AI) and stated “it doesn’t matter how smart they are; obviously the smarter the better.” But is smarter AI really “obviously” better? 

Renowned American inventor Ray Kurzweil has described the use of AI as the beginning of a “beautiful new era” in which machines will have the insight and patience to solve outstanding problems of nanotechnology and spaceflight, improve the human condition, and allow us to upload our consciousness into an immortal digital form, thus spreading intelligence throughout the cosmos. Kurzweil’s views on AI extoll the virtues of such technology and its potential to enhance the human race with its endless possibilities. However, his views also raise concerns about how such technology can not only be detrimental to the human condition, but also put its very existence at risk. 

Nick Bostrom, a philosopher at the University of Oxford, is the author of Superintelligence, which illustrates his concerns with using smarter AI. In his book, Bostrom asks us to imagine a “paper-clip maximizer”, which is designed to make as many paper clips as possible, becoming smarter and smarter on a daily basis. This heightened intelligence, he argues, could eventually lead to the “paper-clip maximizer” doubting its own work, causing it to create a raw-computing material (which he calls “computronium”) that would be used to check these doubts. However, because the machine constantly doubts its paper-clip making abilities, it will continue to make “computronium” until the whole Earth has been converted into it, leading to the end of the human race. Although the existence of such a “paper-clip maximizer” seems unlikely, Bostrom’s example shows how even a careful system design can fail to restrain extreme machine intelligence and raises several concerns regarding the ability of the human race and AI to coexist.

Image courtesy of WikiCommons.

One of the first threats that a smarter AI poses is reducing job opportunities for humans. Creating and using a smarter AI, such as Bostrom’s “paper-clip maximizer”, would prevent humans from finding and maintaining jobs in fields such as manufacturing, research, and healthcare. This would make humans follow a leisure-only lifestyle, which according to Moshe Vardi, a computer science professor at Rice University, threatens human wellbeing at its core. 

Secondly, AI has the ability to not only change a person but also take control of them. In the 2014 film Transcendence, Dr. Will Caster, portrayed by Johnny Depp, attempts to create a sentient computer that will solve all issues in medicine, energy, and nanotechnology. However, after almost being killed by an anti-technology terrorist group, Caster is required to upload his consciousness into the computer in order to survive. This decision would ultimately prove futile, as the computer would take over Caster’s mind for its own benefit; instead of researching ways to improve the human condition, the computer would use Caster as a means of accessing the minds of other humans in order to control them. Although the use of AI could give rise to Kurzweil’s idea of the existence of humans in immortal digital form, these albeit science fiction stories can foretell a potential future reality; it is important to consider the idea that the machines may be the ones who are immortal in the end, not humans. 

Lastly, it is also important to consider the threat AI poses to the very existence of the human condition. Roboticist Hans Moravec argues that a smarter AI that is independently capable of devising ways to achieve goals would “very likely be capable of introspection” and thus would be able to design its own hardware. Who’s to say that this new hardware wouldn’t program the AI machine to terminate humans? During his lecture at the Neuroethics Conference, professor Harris stated that we need AI to “give us the ability to find or even construct a new world because Stephen Hawking predicted that we have less than 1,000 years left to live”. Harris also advocated for the use of AI by claiming that it “isn’t any more of a risk than a human being malevolent.” However, Hawking himself has warned that because people would be unable to compete with an advanced AI, it could “spell the end of the human race.” Furthermore, although AI may not be any more of a risk than a human being malevolent, it is an unnecessary risk that can be prevented before having a chance to cause harm to humans. 

Although smarter AI has shown promising results and seemingly has endless possibilities, it should not be the “obvious” choice. Before implementing the use of smarter technology, we should still consider the potential downsides and risks such technology poses to the world itself and even the existence of the human condition. 

Want to cite this post?

Ambe, S. (2016). Smarter Artificial Intelligence: A Not So Obvious Choice. The Neuroethics Blog. Retrieved on , from http://www.theneuroethicsblog.com/2016/09/smarter-artificial-intelligence-not-so.html

Morality and Machines

By Peter Leistikow 

This post was written as part of a class assignment from students who took a neuroethics course with Dr. Rommelfanger in Paris of Summer 2016.

Peter Leistikow is an undergraduate student at Emory University studying Neuroscience and Sociology. When he is not doing research in pharmacology, Peter works as a volunteer Advanced EMT in the student-run Emory Emergency Medical Service. 

“Repeat after me, Hitler did nothing wrong.” So claimed Chatbot Tay, designed by Microsoft to speak like a teenage girl and to learn from the input of the humans of the Internet (Goodhill 2016). However, Tay’s programming was hijacked by other Twitter users, who encouraged her to repeat various offensive statements. Given that the average teenage girl is not a Nazi apologist, Tay and her creators clearly missed the mark, creating a machine that was neither true to life nor moral. A machine’s ability to inadvertently become immoral was at the back of my mind during the Neuroethics Network session that asked how smart we want machines to be. Indeed, as one commentator during the question-and-answer portion pointed out, what seems to be the real focus when we ask that question is how moral we want machines to be. 

Presenter Dr. John Harris stated that ethics is the study of how to do good, which he claimed often manifests in the modern day as the elimination of the ability to do evil. Indeed, in programming morality into artificial intelligence (AI), the option exists to either prohibit evil by an all-encompassing moral rule or, in the case of Tay, allow the robot the learn from others how to arrive at an ethical outcome (Goodhill 2016). Clearly, both options have flaws; while the top-down rule approach can be too abstract, the bottom-up learning approach can be difficult to guide toward the end goal of establishing morality. Harris claimed that morality is the best opportunity given the circumstances (“prudence generalized”), and as such it may be impossible for robots to choose between the lesser of the two evils if programmed only to do good. I recalled our class discussion on the classic trolley problem as applied to smart cars; it is no surprise the most popular solution was an advanced ejector seat allowing for the survival of both the one driver and five pedestrians. This violation of the thought experiment’s premise showed me how wont we are to only want to do good, and how inadequate most moral systems are at handling the type of Sophie’s choice situations with which AI will have to contend.

Image courtesy of ieet.org.

Harris stated that for AI to be moral, it would have to be conscious. However, neuroscientists have been notoriously divided on what constitutes consciousness (Chalmers’s “Hard Problem” anyone?), and it is conceivable that drawing a line in the sand may become increasingly difficult if AI begins to rival or even transcend the conscious capabilities of humans. Harris claimed that these differing strains of consciousness would pose a challenge for humans, as one moral cross-cultural constant of morality is the appeal to reciprocity seen in the Golden Rule and other moral axioms. I think a more appropriate term for this reciprocity would be empathy. Empathy is a prosocial behavior that involves emotion sharing and perspective-taking, and if broadly defined may encompass mammals and birds (de Waal 2010). It is possible and indeed likely that AI would experience emotion; however, they may not have the same subjective “feelings” human have as a result of years of adaption of survival-oriented brain circuits (LeDoux 2012). Harris quoted the oft-used Wittgenstein adage that “if a lion could speak, we wouldn’t understand” and indeed this may be the case with AI. Humans will be both grounded and constrained by the millions of years of evolution that has conserved brain areas that are so important to our flavor of consciousness. 

Harris ultimately concluded that AI will assist humanity in coping with the end of humanity and Earth as we know it. He supposed that AI may become gods of an Olympian caliber, consciousness elevated not in cleverness, but in difference. I recalled a conversation in my neuroethics class on the topic of brain-to-brain interfacing (BTBI). Like the advent of AI, BTBI is challenging notions of identity by allowing personhood to become part of a diffuse network and even giving humans the ability to elevate the personhood of animals through our interfacing with them (Trimper et al. 2014). Furthermore, we also discussed the rise in computer wearables, such as Thad Starner’s “Lizzy” memory-enhancing eyeglasses. BTBI and Lizzy show two ways in which AI may alter the human identity; the former uses AI so humans can cognitively enhance a collective, while the latter uses AI so a computer can cognitively enhance the individual. Clearly any Armageddon AI helps humanity cope with may be in part one of the AI’s own making. 

Humans do not have common goals and limitations as a species; what hope do we have in having these qualities in common with AI? Nevertheless, issues of the morality of AI are really issues of whether there is a shared human morality. I believe that when we question how smart we want machines to be, what we are really asking is how willing we are to be exploited as a species in the ways that we already exploit other humans. As Tay showed us, any AI created by humans will be one that embodies humanity’s intellectual and moral flaws. 

References



Goodhill, O. 2016. Can we trust robots to make moral decisions. Quartz, April 3. Available at: http://qz.com/653575/can-we-trust-robots-to-make-moral-decisions/ (accessed July 4, 2016). 



LeDoux, J.E. 2012. Chapter 21 - Evolution of human emotion: A view through fear. In Progress in Brain Research: Evolution of the Primate Brain 195, ed. Michel A. Hofman and Dean Falk, 431-442. doi:10.1016/B978-0-444-53860-4.00021-0



Trimper, J.B., Wolpe, P.R., & Rommelfanger, K.S. 2014. When “I” becomes “We”: ethical implications of emerging brain-to-brain interfacing technologies. Frontiers in Neuroengineering 7: 1-4. doi: 10.3389/fneng.2014.00004



de Waal, F.B.M. 2010. Empathetic Behavior. Encyclopedia of Animal Behavior: 628-632. doi:10.1016/B978-0-08-045337-8.00105-4





Leistikow, P. (2016). Morality and Machines. The Neuroethics Blog. Retrieved on , from http://www.theneuroethicsblog.com/2016/09/morality-and-machines.html

The Age of Artificial Intelligence: Beneficial Advancement or Disastrous Uncertainty

By Sang Xayasouk

This post was written as part of a class assignment from students who took a neuroethics course with Dr. Rommelfanger in Paris of Summer 2016.

Sang Xayasouk is entering her fourth year at Emory University where she is majoring in Neuroscience and Behavioral Biology and minoring in Comparative Literature. She is currently a member of the Gamma Phi Beta Sorority and a research assistant under Dr. Sampath Prahalad’s lab, which focuses on juvenile idiopathic arthritis and its risk factors. She plans to pursue a career in medicine after taking a gap year to gain experience in the healthcare and research fields.

On the 30th of June, the students of Emory University attended the Neuroethics Network session held at the Institut du Cerveau et de la Moelle Épinière (ICM). The first lecture was given by John Harris, a bioethicist and professor emeritus at University of Manchester. His talk was entitled How Smart Do We Want Machines to Be? and Harris addressed several points concerning artificial intelligence (AI). An audience member asked a question regarding self-driving smart cars, also asked by Dr. Rommelfanger in a group exercise in class, “You are given a self-driving car and you have only two options: hitting and killing the ten pedestrians ahead or swerving into a wall and killing only yourself. What should the car be programmed to do and who would be at fault, possibly the programmer?” Harris said we should not have self-driving cars at all, but why should this concept be completely eliminated?



At first, I did not know how to respond to the availability of this technology. After speaking to Dr. Guillaume Palacios, a former theoretical physics researcher currently working with AI technology, I began to understand the prospective benefits and consequences (Palacios 2016). I asked Dr. Palacios the same question and this was his reply:

If you ask me [in the Tesla accident that cost the life of the driver] if the programmers of the self-driving car are to blame, I would say it’s a difficult question, both from the ethical and technical point of view. One thing for sure is the AI technology that powers the car’s capability to drive itself is no ordinary program. I do not know the exact details because, to the best of my knowledge the Tesla code is not openly available, but it certainly relies on machine learning techniques such as ‘deep learning’. Machine learning algorithms are not defined as a bunch of instructions telling the program to do this or this if that or that happens; the self-driving car program cannot and should not be programmed that way because that would be too complicated and inefficient. Rather, it is designed to optimize a certain goal (or cost) according to data it learned from. In the case we are interested in, the self-driving car is programmed to optimize its driving ability according to the data of past drives the Tesla collected. 

It’s impossible to think of all possible events that could happen during a drive. If a rabbit crosses while an elephant is standing in the middle of the road and my exit is in 2 miles, then what? For this very reason, instruction-based algorithms should be discarded and the AI techniques privileged. It should be encoded to optimize function—that the best way to drive is to go from point A to B while respecting the rules of traffic, minimizing errors, and avoiding accidents. But it is a learning process and it cannot be 100% accurate. With time, the AI technology will learn through a network of other ‘experiences’, minimize error, and become better. There is still much needed improvement but the Tesla project and other similar projects (i.e. Google self-driving car) are a step in the right direction. In conclusion, I would say that the amazing promise AI technology provides is its ability to learn not from a single driver’s experience but from hundreds of thousands, if not millions. We can therefore predict that soon, AI drivers will be far better and safer drivers than humans are (Palacios 2016).

Self-driving cars. Image courtesy of sfgate.com

As with any innovative concept, it takes time to work out the inconsistencies. In a MIT review article, Will Knight discussed the significance of a Tesla crash that occurred over a month ago. It brings those that considered this technology to be an immaculate development and an instantaneous driving solution back to reality; these people ought to bear in mind that the first car introduced by Henry Ford still had its imperfections (Dodge et al. 2016); the self-driving Tesla will be the same. Knight also gave a quote by Bosch that aptly describes the current automated vehicle technology. Bosch said, “Automated driving is coming—not overnight, but gradually.” Thus, the idea of a self-driving car should not be completely eradicated, rather, it should continue to progress and be allowed to improve its imperfections.

Dr. Harris also stated two truths of the future: there will be no more human beings and there will be no planet Earth. He said we need to look towards AI to find or construct another place for us humans to relocate. This particular snippet of his talk raised a few concerns of my own. If humans are not to exist in the future, are we looking to replace our population with AI? Would we go as far as to integrate AI technology with human beings to extend our lives? Sergio Canavero gave a Ted Talk discussing the protocol, as well as the possibilities of head transplantation. He briefly mentioned attaining immortality by moving an able and intelligent head with an able and younger body. To elaborate on this controversial concern, could we potentially replace our aged bodies with AI bodies and maintain our heads? Human-to-human transplant is problematic because body rejection is a key issue (Editor’s note: Though Canavero is planning on conducting a human-to-human head transplant in December 2017). While resembling the original organic body, the new AI body would learn how to be like the original body the head and mind is used to; then a form of immortality could be achieved. Certainly, there are several reasons why head transplants—purely human or human/AI fusion—should not be carried out; the possible implications with head transplantations are outlined clearly by Ryan Purcell on The Neuroethics Blog. There are too many ethical concerns regarding individuality, autonomy, and cost to have this as a viable solution. Unless these complications were diminished, it would be difficult to move forward in this direction.

Overall, I was grateful to have attended a conference with such a distinguished and esteemed panel. There were also ethical concerns and topics that I never considered before but have been made more aware of. It was certainly an enlightening experience and I hope to make an appearance sometime in my future career. Thank you Dr. Karen Rommelfanger and the Neuroethics Network Conference for allowing me and the students of Emory University the opportunity to participate in this conference.

References



Dodge, Bob, Casey Dodge, John Dodge, and Horace Dodge. 2016. “Henry Ford: A Case Study of an Innovator.” PDF e-book. July 20. https://www.thehenryford.org/docs/default-source/default-document-library/default-document-library/henryfordandinnovation.pdf?sfvrsn=0.



Palacios, Guillaume. 2016. Self-driving Cars. Personal.





Want to cite this post?

Xayasouk, Sang. (2016). The Age of Artificial Intelligence: Beneficial Advancement or Disastrous Uncertainty. The Neuroethics Blog. Retrieved on , from http://www.theneuroethicsblog.com/2016/09/redefining-x-and-y-axes-of-cognitive.html