Hi, Westworld fans! Do you want to go deeper into the science of AI and robot cognition? Could we at some point be living in our own Westworld? We spoke with cybersecurity and tech expert and Westworld fan Avi Chesla, founder and CIO of cybersecurity firm, EMPOW.
We’re going to discuss myths and realities with regard to Westworld and general misconceptions, how machine learning can mimic or even replicate human behavior, the implications of Westworld on the cybersecurity community, and more. Join us for a chat on the robot uprising from a science and tech point of view!
Q: Hi Avi! As a tech geek myself, I’m thrilled to have your POV on the show! What first pulled you into Westworld? What was the defining moment for you?
"A: “Being submerged in the world of AI, it was a natural attraction to the show. Of course the show is strictly fictional—but it is interesting for me to see if the writers have done their homework or not, regarding artificial intelligence, robotics, and machine learning. It actually took me till [Season 2] to realize that I am a true fan. My defining moment was not when Dolores nor Bernard gained cognition, but the tests that were being performed on Delos. It intrigued me that William and the team could not seem to understand why robot Delos malfunctioned when a question was asked that he did not learn the response to and could not improvise. That’s the defining characteristic of true AI—it doesn’t need embedded knowledge.”"
Q: Westworld is based in the future. It’s odd that Westworld brings up the long-debunked bicameral mind theory given its time frame—what does that tell us about Westworld’s likely origins and its creators?
"A: “The show is based on the 1976 movie Futureworld [and the original Westworld, its precursor] written and directed by Michael Crichton—the same man who wrote the Jurassic Park book series. So it actually makes sense that the bicameral mind theory is being used in the show because the 1970s is when this theory came about. The term was first introduced in a book, The Origin of Consciousness in the Breakdown of the Bicameral Mind. The book, for all its intellectual pretension, became a bestseller of sorts, and I think the writers of Westworld are having some ironic fun here. The creators took advantage of the fact that the original movie consequences—which today we might dismiss as immoral—can easily be replicated and made more believable today. But the underlying premise remains unchanged: a western theme park, rich guests torment the servile robots until they rebel and start an uprising. What is different in the show is the richer texture—the way the writers give each main character their own backstory, and how they overlap to build dramatic interest.”"
Q: How far along are we in replicating the human mind?
"A: “Let me put it like this: Airplanes failed to fly as long as humans tried to make them fly like birds do. The same goes for artificial intelligence, we cannot replicate it completely but we can develop alternative approaches and methods to reach results comparable to what the human brain can achieve. But remember: only similar, not the same level of results as of yet. We are still very far from getting all the way there. AI is still not successful in transforming adapted knowledge and thought processes from one area of expertise (one problem domain) into another. For example, AI that is trained in one environment—such as chess—cannot solve other problems. AI can learn to be proficient at chess but fail miserably at checkers, backgammon, and truth or dare. It can apply rules but not learn and apply new ones, on its own, through neural associations which have been created over time. I believe that true intelligence must involve biological elements from the human brain.”"
Q: What does “human” mean to you? Does biology define us as real, or would AI be just as valid as we are in terms of humanity? How important is biology, taking into account “artificial” organs, bionics, prostheses, etc., and why? For example, someone with a bionic appendage, an implanted spinal neurostimulator (like my sister), or even someone taking medications that affect brain chemistry?
"A: “Biology does define and create our humanity. Our DNA carries thousands of years of knowledge (‘memories’) which computers do not. Humans have evolved on their own, creating highly intellectual beings that are so advanced that we’re the only of our kind. AI/robotics/machine learning can only have purposes that we give it.”"
Q: What’s the biggest difference between a human mind and our current level of AI cognition?
"A: “The application of learning; as mentioned before, humans are doing that all the time, without realizing it. We export associations from one learned field to another, which creates and accelerates learning, which AI cannot do now but maybe will. This is also related to the way our memories and ‘mental filing systems’ have evolved over time. It’s also related to the ‘DNA’ memory I mentioned before. Some things are ingrained in our DNA, and we do not have the means to replicate that yet with AI.”"
Q: What is the next big hurdle we are facing in our journey to AI consciousness?
"A: “Consciousness is a big word—and it relates to the so-called ‘theory of mind.’ There is no single definition for the word, and while some think it will emerge as technology develops, I don’t believe so. That’s because AI and data science are not built for the creation of consciousness, they are built to apply what exists rather than develop what doesn’t. Creativity and inventiveness, however, don’t emerge from logic and calculations, so how can computers achieve that? I believe that consciousness requires a self-organizing system, like the brain’s physical structure. But existing machines are not designed to be as self-organizing as the brain. We cannot achieve “consciousness” until we know how to involve some biological elements in computing. Possibly cloning is the gateway into it? Perhaps a singularity between man and machine will become possible, where DNA and quantum computing fuse.”"
Q: What are some common misconceptions about machine learning?
"A: “There is actually no learning going on. ML cannot, practically, transfer one problem domain adapted thought process to another. It cannot reason. For example, machine learning cannot conduct a cause-and-effect analysis. As a matter of fact, AI cannot really understand by itself if cause and effect are reversible or not. For example, if A leads to B, is it possible that B can lead to A as well? Practical commercial effective solutions always involve embedded expert human rules (heuristics) as part of the AI solution.”"
Q: What are some things about AI that Westworld gets wrong?
"A: “It’s less about getting things wrong and more about leaving things open-ended because the science is not there—particularly in the areas of reasoning and purpose. The writers cannot explain how the hosts arrived at what they are calling consciousness—only that Arnold had something to do with it. The hosts were coded to do only one thing: stay in their loops they were taught. But how did they move from one purpose to another?That cannot be taught; that is biological evolution.”"
Q: What does it get right?
"A: “What Westworld gets right is the idea of a pyramid construct. In the show they simplify the determinants of consciousness, with memory being the basis, then improvisation, followed by self-interest, leaving the top of the pyramid empty for consciousness. I think that they wanted to leave it a mystery because what does it really mean to be conscious? It is posited that our subconscious takes up most of the human brain then followed by the unconscious self and a small portion being conscious.”"
Q: Is it possible for one machine to control another without human prompting at any stage of the process, including initiation?
"A: “Technically it is possible, but only if it is programmed to do so by a human (expert rules are embedded). So it starts with us, with flesh and blood—a human needs to give the machine the idea. AI is not in the stage of self-awareness and understanding to know who should control whom…yet.”"
Q: Would AI be able to evolve on its own by having self-awareness and cognition without a human component to learn from, including proxies?
"A: “Not now. Human (biological) components needs to be involved in order to achieve that part. If we find a way to have a hybrid machine that includes the right biological elements—like brain cells—then it will be possible. Until we figure that component out, machines would not be able to evolve on their own.”"
Q: Can machines only mimic human behavior, or can they adapt it into their functioning, essentially replicating it?
"A: “For now they can mimic some of the basic human behavior. Existing scope of AI doesn’t allow adapting human behavior. But as human behavior adapts, AI can mimic that evolution.”"
Q: What is a cybersecurity measure you would have implemented before or during the robot takeover to avoid or stop it? Or do we need full control over the human mind before we can fully control a replica?
"A: “A robot ‘takeover’ can only take place when self-awareness and self-organizing AI will be a reality. So chill—we are far away from achieving it. Today, if a robot does something we didn’t program it to do, that is only because the AI behind it didn’t work. That’s a design issue, not a robot-on-the-loose issue. The best cybersecurity measure will be to enforce policies that set the parameters of behavior. If the robot breaches these human-defined limits, it should be shut down.”"
Q: What are the biggest takeaways for the cybersecurity industry from Westworld?
"A: “Technology is moving very fast, and we need to be one step ahead of those innovations, and be more involved with the underlying industries. For example; recently there has been a hack in pacemakers. Cybersecurity experts should be more involved at the front end of their development, not at the back end when the design is fully backed. Any device that is hackable can be programed to mass destruct. These devices should be engineered with a solution—already programed to kick in on its own—to fight that and learn for the next attack. There are too many entry points for the dark web to allow for after-the-fact protection.”"
Q: The overarching theme was that reality is not defined by human biology. They found self-awareness in AI by simplifying, not expanding. Do you think this will be the case in reality, or will it just result in robots for show? What expansions are necessary?
"A: “As I argued before, I believe that self-awareness will be achieved only if biological components will become part of the machine. In order to expand you must start with simplification. For right now, robots are for show. They are given commands and execute on them. They learn as well as instruct them. Even when we contradict ourselves, they still continue their programing. Necessary expansions will come from biology, and integrating the ‘must-haves’ of both worlds so that robots can move to the next level, where we can have a further debate about what constitutes consciousness. I don’t see the debates ever ending—in fact they will get more fascinating.”"
Q: If you had to pick three things as a test for true self-awareness and consciousness, what would they be?
"A: “They would be:1. The ability to regret or be remorseful.2. The ability to develop levels of respect toward others.3. The ability to separate oneself from the external world.”"
Q: If you had to pick one character who is/was secretly a host, whom would you choose and why?
"A: “Based on the second season’s finale, I would definitely say that William is secretly a host. In an episode I noticed that when William got wounded in his arm, he reached behind the bar the get a plasma healing torch that is usually used on the robots to repair them. He used the device on himself which we never saw that done on a ‘guest’ before this time—although there are methods of healing that could also support this. And the last few minutes of Season 2’s episode leaves us wondering…”"
Q: For season 3, whose backstory do you want to see told? Whose do you think is most important? Why?
"A: “For season 3 I would love to see William’s daughter, Emily’s [Grace’s], backstory told. We are left at the end of season 2 with William returning to a room, similar to the one where he experimented with Delos, and it seems that roles had reversed. William essentially became Delos, the experiment, leaving the viewer to wonder how long/if even William is/has been a robot. And Emily took on William’s role as the teacher, in regards to the Delos scenario. In both instances the people/robots doing the experimenting, and being experimented on parallels Dolores teaching Bernard all the things Arnold would do as she remembered him before his suicide. It is quite a mind game! I think Ford’s story is important as well. You get a glimpse of Ford’s past with the young robot boy he made of himself, and the home he said he grew up in. I think that it is essential to know Ford’s progression to his ultimate fate, and to figuring out the source of William’s obsession with immortality and why he thinks the game is for himself.”"
Q: How has the rise in medical implementation of things like spinal neurotransmitters and bionics contributed to cybersecurity and the future of AI?
"A: “Learning the body’s communication channels, and how they transport signals to and from the brain, is essential for understanding how to integrate biological components with computerized ones. Spinal neurotransmitters and bionics solutions and research contribute to that understanding—and as I said, self-awareness can be only achieved through the hybrid approach.”"
Q: What twist would you throw in Episode 1 of Season 3?
"A: “Ford comes back as a host after the apocalypse and has achieved immortality.”"
Q: What was the most important moment of Season 2 for you?
"A: “When Akecheta, host in the Ghost Nation, told his story. He found his way through the ‘maze’ with limited human/guest interaction. It showed viewers that even with little exposure to the outside world, these robots could craft their own narratives based on their experiences and direct human prompting.”"
Q: Are there any unanswered questions you think those not in the tech industry haven’t considered?
"A: “I think questions that haven’t been answered are:"
"What/who prompted the hosts, like Dolores, to gain cognition?"
"Why now, after Arnold’s death, 34 years after? What took so long?”"
Q: How do you think machine learning will impact the various hosts’ assimilation into their new world and their interactions with humans in season 3?
"A: “Depends on the host. Dolores has been in the real world and remembers it vividly. The others, not so much…but she can teach them. This will be a real test of machine learning and adaption as the series evolves.”"
Q: What is one aspect of AI cognition the show hasn’t addressed yet?
"A: “The differentiation between dreams and memories. If the hosts have true consciousness they would be able to self-organize these thoughts. This is even hard for human beings at times—neuroscience shows us that we have many false memories and they are hard to correct.”"
Q: Fans want to know: Is a robot uprising possible? Is it probable? What timeframe can we expect for a potential “AI war”?
"A: Once we understand how to really achieve integration of biological brain elements that are compatible with the computerized ones, then we can expect robots to essentially learn human behavior and become more powerful than the human brain, leading to a robot uprising. Consider that human brains operate autonomously and individually—we are not connected to others. But robot intelligence can be amplified into a whole that is greater than the sum of its robot parts.If I needed to predict when, it will take at least 30-40 years from now.”"
Thank you so much for your contribution to our Westworld community discussion, Avi! While we slightly disagree on a few things—such as biology defining us as human and where we draw the line—I think you’re spot-on! We appreciate you answering our questions and we would love to have you back any time.
Fans, tell us what you think about these questions and answers and where they could lead in season 3 and beyond.
Stay tuned to Beyond Westworld as we bring you more articles, discussions, and interviews leading up to season 3!