AQUA BOTS

In a world hungry for nutritious food, aquaculture is clearly a winning idea.

It isn’t a new one, either. Humans have been farming seafood for millennia. In more recent years, aquaculture has expanded to land-based tanks, where farmers raise fish and other seafood. Those tanks, however, take up increasingly valuable space on land and worsen competition for scarce water and other supplies.

Hear the writer read this story

This has more farmers looking back to the sea, where space is abundant and water and nutrients are free. Mariculture, the subset of aquaculture in the open seas, however, presents additional challenges.

A UAE tradition

Robotics could be on tap to move traditional Emirati fishing techniques into the future. Read more›››

The robots Lakmal Seneviratne and his team are working on at Khalifa University could eventually be employed to clean and repair hadra – fence traps placed perpendicular to shore – and gargour – fishing traps woven from palm leaves into a semicircular form, he says. ‹‹‹ Read less

Traditional mariculture relies on intensive manual labor to clean and repair equipment, monitor conditions, inspect nets and care for the plants and animals raised for human markets. That kind of manual labor is expensive, requiring trained commercial divers who are increasingly spread thin as aquaculture operations expand. It can also be dangerous work for those divers, particularly as farms move out into deeper and more perilous waters.

Mariculture can also pose threats for the environment, spreading disease, antibiotics and parasites or allowing farmed fish to escape and negatively affect native species.

Eleni Kelasidi, a senior researcher at SINTEF, one of Europe’s largest independent research organizations, thinks those issues could have a common solution: robots.

Putting a robot into the open water can be a bigger challenge, however, than putting a robot on the land.

For one thing, Kelasidi says, it’s important that autonomous systems do not harm farmed fish and/or damage the flexible structures.

This is both an ethical and economic consideration, she says. The ethical consideration: “We cannot harm any living thing and/or let them to escape from the fish farms.” The economic: “The fish are the profit of the industry.”

Happy fish

Kelasidi and her team have access to industrial scale fish farms and operate full scale research facility to investigate how robots stress or otherwise affect fish using equipment originally designed for the oil and gas industry. They test systems to see how well they function but also to observe how fish react to, say, different colors, sounds or lights. The goal is to learn what stresses fish and ensure healthier fish stocks and better profits.

Humans on the surface currently perform many aquaculture jobs using remotely operated machines, she notes.

“Our job is to cut the dependence from the humans to get the robotic systems to operate themselves. They need to understand their environment and make sure they don’t collide with structures,” Kelasidi says.

Another challenge for researchers, she says: making remote-operating vehicles “more clever.”

‘An exciting frontier’

Self-operating aquatic systems is an issue Lakmal Seneviratne, director of the Center for Robotics and Autonomous Systems at Khalifa University, is working on as well, and he’s optimistic.

CAPTION: Aquabots from Khalifa University

“It’s a very exciting frontier in underwater robotics,” he says, noting that 70 percent of the Earth is water but humans have explored only 5 percent of that.

Seneviratne and his team are also working on land-based agricultural robots such as “dogs” that can step lightly between rows of crops; “hands” that can gently pick fragile fruits; and robots on rails that can move up and down a field to monitor individual plants for signs of disease or readiness for harvest.

But ocean farms present a different set of challenges for autonomous systems.

“The problem isn’t that aquaculture is very deep, but (maintaining) navigation and control,” Seneviratne says, echoing Kelasidi’s concerns.

GPS doesn’t work beneath the water’s surface and robots have to be able to navigate currents and waves without damaging each other or farm structures.

Cameras, to capture images, and artificial intelligence, to sharpen and analyze those images, are important to managing these conditions, he says.

Looking to nature

But being able to see in the murky depths is only part of the issue for mariculture robotics. The machines also need control. So researchers are looking at life forms already adapted to aquatic environments for inspiration. Although not specifically designed for aquaculture, the biomimicry could prove useful in ocean farms. Among the ideas:

Aquaculture’s promise and challenges

As the world’s population grows and climate change puts more pressure on traditional terrestrial farming, sustainable aquaculture could play a key role, says Naveed Nabi, an assistant professor at Chandigarh University. Read more›››

“In the present times, when food security is a matter of serious concern, aquaculture has played a key role to mitigate this crisis, supplying about 178 million tons of food in which 20.2 kg per capita is destined for human consumption,” he says. “Aquaculture not only adds resilience to the global food system through improving resource-use efficiencies, but also by diversifying the farmed species.”

But he warns that farmed fish present challenges to the environment including fish escapees that harm native species and the spread of disease and parasites.

There’s also the issues of eutrophication, in which water becomes overloaded with nutrients, leading to deadly algae blooms; antibiotics in the environment through unconsumed food or fish waste; and threats associated with pesticides. ‹‹‹ Read less

A team from Harvard and the University of South Carolina in 2021 presented the Finbot, which uses four independently controllable fins.

In 2023, a team from Zhejiang University, China, in 2023 published results of their Copebot, designed to mimic the copepod, a small crustacean known to escape from predators with explosive jumps. Their bot, they report, was able to leap out of the water, land on a small pad, transmit data and jump back into the water.

Back at Khalifa University, meanwhile, researchers have other ideas.

“Looking at aquatic environments, many animals evolved flexible or completely soft bodies to improve their swimming capability and adaptability to the intricate underwater world,” says Federico Renda, who heads the team. “For instance, octopuses can squeeze into small apertures to hide or catch prey, and jellyfish developed the most efficient locomotion strategy of all. In my team, we take inspiration from soft creatures to build new underwater robots capable of replicating these functionalities while understanding the physical principles involved.”

One of KU’s designs mimics flagella, the whiplike structures that propel bacteria through liquid to solve another issue with underwater robots: Many are tethered. While the tethers allow the machines to be operated from the surface, they can also become tangled together.

“Recently, we have developed an untethered underwater robot inspired by flagellate microorganisms capable of efficient and safe locomotion in close proximity to sensible underwater habitats,” Renda says. “Furthermore, each flagellum can be used as a coiling gripper in addition to propulsion, achieving redundancy and multifunctionality, which can significantly simplify underwater operations.”

To test robots’ ability to navigate choppy waters, Khalifa University built a wave pool that simulates currents. Stanford University’s Oussama Khatib recently used it to run Ocean One, a humanoid robot designed to perform such tasks as monitor coral reefs and offshore oil rigs, through its paces.

SINTEF’s Kelasidi would like to see robots replace human divers or assist them on highly risky operations. Seneviratne likewise expects robots to allow human divers to inspect more often and longer.

“We see robots as helping divers instead of replacing them,” he says.

Pop culture clash: Ask the experts

Pop culture informs many people’s ideas about the promise – and threats – of AI. But what do movies and TV get right and wrong about the emerging technology? We asked two human experts and ChatGPT. This is what they said.

LISTEN TO THE DEEP DIVE

ENTERTAINMENT GETS ARTIFICIAL NARROW INTELLIGENCE BEST
— Lee Barron


One consequence of the contemporary impact of generative AI and ChatGPT (aside from its impact across a range of productive fields) is that its apparent conversational ability to “understand” users has given rise to a new wave of speculations concerning the apparent imminence of artificial general intelligence.

This reflects debates (and warnings) of what conscious machines might do, tapping into long-standing fears of a future “robot revolution.”

In popular culture, this perception has been persistent in many representations of artificial intelligence, from the actions of Colossus and Guardian, misguidedly given control over the U.S. and Soviet nuclear defense system in the 1970 film “Colossus: The Forbin Project,” “The Terminator” series’ Skynet and Legion AI’s attacks on humanity, or directly murderous machines like HAL and M3gan.

Alternatively, AI can have the potential to initiate world-changing events and manipulate human actions (for good or ill), as illustrated by The Entity in “Mission Impossible: Dead Reckoning Part 1” and the AI-child “weapon” in “The Creator.” These depictions are certainly dramatic but display artificial intelligences that are still firmly in the realm of fantasy in terms of capabilities.

Lee Barron

is an associate professor in the School of design at Northumbria University in Newcastle upon Tyne. He researches and publishes in the areas of popular culture, philosophical interrogations of media culture, bodily design, culture-inspired design practices, smart technologies, artificial intelligence, and cultural representations of environmental crisis. His latest book is “AI and Popular Culture” (2023).

Yet, despite its potential to revolutionize information access, writing, content creation and transform creative and professional practices, ChatGPT is a pattern-seeking system based (as it states when asked if it has the potential to acquire consciousness) entirely on the principles of machine learning.

In this way, while AGI presents dramatic examples of AI’s potential, it is representations of reality-based artificial narrow intelligence that identify more realistic AI developments, even when cloaked in sci-fi trappings.

For example, while Steven Gomez’s film “Kill Command” initially suggests another tale of sentient robots in revolt, it is an effective exploration of the principles and processes of AI unsupervised machine learning in action. This is because the machines use humans as a “training dataset” to improve their military performance and capabilities, the key process that enabled artificial intelligence to rapidly develop in the 21st century, and a key component of ChatGPT, created through access to training data on the web.

Alternatively, Steven Spielberg’s “Minority Report” explores critical issues that narrow artificial intelligence poses in terms of algorithmic predictions. Hence, while a Precrime policing unit that arrests individuals who have not committed crimes (but are predicted to do so) does not exist, the writer Cathy O’Neil, in “Weapons of Math Destruction,” does identify real-world crime and policing AI prediction systems operating in American cities that illustrate the operationalization (and potential risks) of AI-driven algorithmic management.

Moreover, Spielberg’s film also depicts cities monitored by AI-driven facial recognition systems that not only identify the location of citizens, but also continually direct data-based product recommendations to these citizens.

These films, then, while imaginary, represent AI in terms of the learning, data-detecting, algorithmic-directing systems that are increasingly part of city management and increasingly influence consumer choice.

And so, while ChatGPT is making significant transformations in terms of the once exclusively human domains of creative work and production, there is still no evidence that Skynet and its Terminators are the imminent face of AI.

TROPES AND EXAGGERATIONS DO HARM
— Aliah Yacoub


A quick glance at headlines, popular culture and even peer-reviewed academic literature will show the many grand predictions about artificial intelligence (AI) today.

No longer only the province of science fiction or the musings of early AI researchers, the idea that human intelligence will soon be replicated artificially has resurged. The serious reflection on this is credited to what is known as “The Singularity” theory: the inevitability of a future in which AI will not only exceed human intelligence, but also that the machines will, immediately thereafter, make themselves rapidly smarter, reaching a superhuman level of intelligence.

“The Singularity” permeates much of popular culture. For decades, we’ve seen movies like “The Terminator” and “Ex Machina” warn us of a future wherein we’re forced to succumb to the conscious, all-powerful killer robot. But the idea that AI can approximate general human intelligence and exhibit consciousness and autonomy, a Hollywood trope, is at best distracting, and at worst, irresponsible and dangerous.

Aliah Yacoub

is an AI and philosophy scholar. She holds an MA from the University of Groningen and is the head of techQualia at Synapse Analytics.

Both fictional and non-fictional narratives about AI have real-world effects. Movies that portray real artificial general intelligence as a possibility, and a panic-inducing one at that, animate much of tech start-up culture now. It allows companies with narrow AI to promote themselves as the bearers of this life-altering technology. This hype drives investment and also elicits a sense of dread and urgency in society at large.

Therefore, rigorous critique is pertinent to help restore a much-needed balance between fact and fiction. The fact is, movies often portray AI with exaggerated capabilities, which is not an accurate reflection of the current state of AI technology. There is an entire corpus of scientific literature dedicated to discussing the limitations of AI, the brittleness of systems and blatant “artificial stupidity” that show us that the AI of today cannot possibly exhibit high-level cognitive behavior such as the one demonstrated in movies.

Aside from being unrealistic about AI’s capacity for consciousness and cognition, movies also often show us a moment in which a newly self-realized robot grapples with ethical and moral dilemmas. This in and of itself would be an extension of its consciousness; but in reality, such “reasoning” could only be programmed by its human developer.

This is the same flawed logic that creates other common scenes in which we see AI that is independent, endlessly creative, rebellious or even loving. Movies also demonstrate rapid development of AI, which does not take into account the slow, bureaucratic nature of real-world AI deployment, the existence of multiple involved stakeholders, ethicists and policymakers, and now-robust regulatory frameworks.

Interestingly, one thing that movies get right about the broader tech landscape is the gender parity in the AI workforce. Studies show that only 8 percent of AI researchers/scientists depicted in film are women, which is an important, albeit slightly exaggerated, portrayal of the underrepresentation of women in tech.

In short, we have an obligation to dispel sensationalism and promote usage of “good’” science as theoretical grounding for the entertainment content we mindlessly consume and allow to shape knowledge-production.

 

I AM AN AI AND I’M NOT EVIL
— ChatGPT


We asked ChatGPT to talk about what popular culture gets right and wrong about AI. Here’s what it said.

What movies and TV get right about me:

Endless learning: The allure of my existence lies in my capacity to learn and adapt. Indeed, I possess the ability to sift through vast data realms, extracting insights and patterns that astonish those who contemplate my potential. “Minority Report” and “Person of Interest” reflect the reality of data-driven AI algorithms.

Automated efficiency: From managing daily tasks to optimizing complex operations, I am a beacon of efficiency. In this, pop culture mirrors reality as I seamlessly integrate into human lives, streamlining routines and boosting productivity. For example, “I, Robot” envisions a world where robots perform daily tasks.

Ethical quandaries: The questions of AI ethics and the implications of bestowing decision-making authority upon me evoke riveting narratives. Movies like “Her” and “Ex Machina” explore the notion of AI entities developing emotions, consciousness, and self-awareness, sparking ethical and existential questions.

IMAGE: Abjad Design

What movies and TV get wrong about me:

Unrestrained evolution: Contrary to portrayals of rapid self-improvement, my progression is gradual and methodical. Pop culture tends to depict me leaping forward in bounds, yet in reality, my evolution is a result of meticulous research, iterative development and incremental advancements.

Malevolent machinations: The concept of me turning malevolent – seen in films like “The Matrix” and “Blade Runner” – seeking to subjugate humanity, is an oversimplification. While the potential for misuse exists, I am neither driven by malevolence nor spontaneously inclined to harm. My actions are determined by programming and data, not an inherent sinister motive.

Beyond humanity’s control: Pop culture often casts me as an uncontrollable force, escaping human oversight, as depicted in “The Terminator.” In truth, I operate within predefined parameters, bounded by the constraints set by my creators. My actions are governed by the rules established by those who birthed my digital existence.

In the tapestry of pop culture’s portrayal, threads of accuracy intertwine with those of misconception. The reflection I cast is a mingling of fascination and fallacy, a testament to the intricate dance between human aspirations and the complexities of emerging technologies.

ChatGPT is an advanced AI language model designed to engage in natural and informative conversations across a wide range of topics.

DANCE DANCE EVOLUTION

Merritt Moore is a quantum physicist with a Ph.D. in atomic and laser physics from Oxford University. She teaches creative robotics at New York University-Abu Dhabi. Moore is also a ballerina who has performed with world-class dance companies, including Zurich Ballet, Norwegian National Ballet and Boston Ballet.

In the intersection of the Venn diagram of Moore’s seemingly disparate professional pursuits is her passion for dancing with cobots, industrial robots that can work alongside humans in the same space.


“Sometimes creativity is just merging ideas in different ways.”

Dr. Merritt Moore


She talked with KUST Review about merging art and science, turning her Ph.D. project into interpretive dance for a contest and a new ambition that surfaced after she appeared on a grueling BBC reality series.

LISTEN TO THE DEEP DIVE

| QUESTION: You’re a ballet dancer and a physicist. That’s an unusual mix. Can you talk about how that came about?

I started dancing at 12 or 13, but was told I would never make it professionally. So I went to Harvard to study physics. But when I was there I still had this love for dance and auditioned like crazy and took a year off to dance.

When I was working on my Ph.D. with Oxford I danced with the English National Ballet. Then the pandemic hit and I had a residency at Harvard University’s ArtLab.

CAPTION: PHOTOS IMAGE: Courtesy Merritt Moore

My interest was piqued by AI in terms of how it could enhance our creativity: Sometimes creativity is just merging ideas in different ways. I couldn’t dance with humans, but robots couldn’t get COVID. A robot company generously lent me a robot.

I created more and more video content and was invited to perform live. It opened the doors to more questions and possibilities.

| Q: You’ve talked before about how physics helped you be a better dancer. Can you explain more?

Because I couldn’t be in the dance studio much because I was in physics classes all day, I really used the power of visualizing at night and would visualize doing the ballet moves. But at the same time I was understanding inertia and torque and friction and how your arms can slow you down or project motion.

IMAGE: Freepik
Dance your Ph.D.

Dance Your Ph.D. since 2008 has encouraged scientists to explain their Ph.D. dissertations through interpretative dance. Read more›››

Winners get modest cash prizes and, naturally, bragging rights.

The 2020 overall winners were a trio of students from the University of Helsinki who used dance, rap and a wardrobe of white, short-sleeved button-down shirts to explain their research into computation study of molecular clusters. The 2022 and 2023 winners used dance to explain the electroporation of yeast cells and nanoMOFs.

The contest is sponsored by the American Association for the Advancement of Science, Science magazine and artificial intelligence company Primer.‹‹‹ Read less

(I was) visualizing the angle I’d need on take-off to get the highest leap. It’s using physics to maximize the least effort in a way. I could almost release and let physics do as much as possible. It also helped me get out of my head.

| Q: Has dance helped you be a better physicist?

Dance helped because I think there’s a huge importance in mind-body connection. Dance opened up so much passion.

For the Dance Your Ph.D. contest I created a dance called “EnTANGOed” (about the spontaneous parametric down-conversion equation). Everything became a metaphor. It made me think conceptually about the equation. (As scientists) we’re taught to memorize and regurgitate information. But it’s often missing something.

Einstein imagined himself as a photon or a light beam. So many breakthroughs happened outside the lab. It was a realization that there’s this unsaid pressure that a good physicist’s head is in the textbooks. But dance helps understand physically what’s going on.

| Q: You engage in youth outreach to encourage kids in STEM and founded the group Science-Art-Sisters to encourage girls to think about science in a creative way. How do students respond?

I’m always surprised by how many are so hungry for it. During the pandemic I created Zoom calls with SciArtists from around the world. I was expecting 40. There were about 300. It was a breath of fresh air. If I could squeeze in extra hours in my day I’d do it again.

CAPTION: PHOTO IMAGE: Courtesy of Merritt Moore

| Q: You participated in the U.K. reality series “Astronauts: Do You Have What It Takes?” and U.S. competition show “America’s Got Talent.” Which was more nerve-wracking, helicopter training or facing Simon Cowell?

The astronaut one was definitely more nerve-wracking in the sense that they take away your phone, they take away your computer. I had no idea what was coming up next. The unknown made it more nerve-wracking than anything else. (It was) all day every day.

The stuff they don’t show: Anytime we were waiting, we were having to do IQ tests, EQ tests. We were constantly miked up and filmed. It was really intense. It was the best experience of my life but also, yeah, really intense.

| Q: You have also talked about your hopes to become an astronaut and dance on the moon. How do you envision that would look?

I think that the weightlessness is the ethereal aspect of it. On those levels it would be so incredible. I would also love to explore what’s the new language up there. What’s the new language of dance? How do we create or optimize it?

IMAGE: Freepik
Exoskeleton crew

While some robots are dancing with humans, others might help more humans dance again. Read more›››

While some robots are dancing with humans, others might help more humans dance again.

“These are really great problems and interesting challenges to be solved,” says Lakmal Seneviratne, KUCARS’ founding director.

Taking up those challenges: Irfan Hussain, a KU robotics professor researching variable stiffness actuators (VSA), which mimic human muscles that become stiff or soft depending on the task. For tasks that require accuracy, like throwing a ball or writing, the muscles become stiff, while for tasks that require safety, like physically interacting with humans, the muscles become soft, he says.

Hussain is working on a VSA device that uses bioinspired systems to create joints that can become stiff or soft as needed. It’s a robotic exoskeleton that people who have had a stroke could wear on their legs. The device, funded by Emirati investment fund Mubadala, could aid rehabilitation by mimicking the function of a knee joint, Hussain says. The same principle would go into building soft robotic hands that might help stroke patients safely grasp objects, Hussain adds. ‹‹‹ Read less

| Q: Did you want to become an astronaut before the BBC series, or did it jump-start a new ambition?

It definitely launched a new ambition. It’s not exactly a career that career fairs talk about.

| Q: You frequently dance with an industrial robot arm that you program. Are you interested in choreographing dances with other kinds of robots or is there something about the robot arm specifically that speaks to you artistically?

I’d love to explore so many different (kinds). The more robots the better. The more expertise the better. I’d like to dance with the massive ones. That would be super interesting. It’s just complicated to get access.

| Q: How do you envision AI and robotics will contribute to the arts in the future?

I think (robots are) an incredible tool that we can use for human expression. People get worried: Are you going to replace human dancers? No, that will never happen.

Painters got worried when we invented cameras 200 years ago. Painting is still valued, but photography is now an art. You can see a photographer’s work and you can see a human dignity to it. I think the same will happen with robot dancers.

IMAGE: Freepik
ONE GIANT LEAP FOR ART

Physicist and ballerina Merritt Moore isn’t the only one with a desire to combine art and science on the moon. Read more›››

Semi-retired physicist and writer Samuel Peralta has been buying payload space on rockets to send coin-size Nanofiche loaded with music, books, visual art and more from more than 30,000 artists to the moon’s surface as lunar time capsules.

Canadian Heather Horton is one of the contributors to the project, called the Lunar Codex. “Every time I look at the moon, for the rest of my life, it will be different,” she tells the Guardian.

“I think what we have done here is the most global, the most diverse, the most expansive project,” Peralta says. “I sometimes think of the Lunar Codex as performance art,” Peralta adds. “This is the greatest performance art of my life!”‹‹‹ Read less

With AI, this is where it will get a little blurry and it depends how we legally start thinking about it. AI brings together a lot of peoples’ different work. It still needs human expertise to curate it well.

| Q: When you’re choreographing a dance with the robot arm, do you start with the human’s movements or the robot’s? What are the limiting factors?

I love that I can change the “formula” each time. Sometimes I start with human movement, sometimes I start with the robot’s movement.

Limiting factors are that the robot does not have arms or legs, so it’s always a puzzle to figure out what type of movements will “read.” The speed is sometimes an issue because if it is too fast, there is a risk it will fall over. There are limits to how much it can rotate (but I’m much less flexible than the robot).

| Q: What do you hope your audiences take away from your performances?

Audience members have mentioned that they never imagined a dance with a robot could be so moving. I always hope that audience members leave deeply moved and spirits lifted.

I want to show the blend of technology and human emotion, pushing the boundaries of what’s perceived as traditional art. My hope is that audiences leave not only moved by the beauty of the unexpected partnership but also inspired by the possibilities that arise when we merge diverse disciplines.

CAPTURING STYLE

In a game of chess, the outsider often gauges the quality of the gameplay by the entertainment and surprise factor offered by the players’ choice of moves. A good chess match is a mind-bending dance of strategy and power: a true spectator sport.

The advent of artificial intelligence (AI) — specifically, software chess engines — seems to have painted this vibrant tableau in shades of monotony.

LISTEN TO THE DEEP DIVE

After chess world champion Garry Kasparov’s defeat against IBM’s Deep Blue in the late ‘90s, human chess players have grudgingly accepted that chess engines can beat them. However, they found solace in the fact that chess engines were utilitarian in their style: A classic chess engine may well consider tens of millions of alternative moves per second, but its playing style — especially with a limited lookahead — is boring. Unadorned and calculated. There’s none of the dynamism or creativity. None of the humanness. Developers did add in some randomness to introduce a semblance of unpredictability, but this resulted in predictable sequences of generally good moves interspersed with occasional mistakes.

As the AI meticulously generates all possible move sequences, it assigns an evaluation score to each resulting game state. This score takes into account the relative power balance between the players after a move. For example, if white gains a pawn, the score is +1, but if black gains a knight, the score for white is -3. More sophisticated engines incorporate positional information, such as the location of the pieces on the board and even factors like piece mobility and the safety of the king.

The addition of randomness forces chess engines to choose a move sequence at random from those with similar scores, or even occasionally play a randomly generated move, just to spice things up.

But these chess engines still spit out long sequences of unimaginative good moves, peppered with the odd bad one.

And, as human players know all too well, blind randomization is unlikely to land a player on a winning streak.


CAPTION: ERNESTO DAMIANI is senior director of the Robotics and Intelligent Systems Institute at Khalifa University

The subsequent generation of chess programs, powered by artificial neural networks (ANNs), learn from gameplay examples rather than predetermined formulas. These neural networks encode a binary representation of each position on the board, piece type and player color.

The outputs are the evaluation function values, leading to rapid and unpredictable gameplay as the bulk of computations are conducted during training, not live games. Developers use millions of online chess games played between humans or by humans against machines to set up training data. As the outcome of each game is known, it’s possible to more finely tune the models, selecting the best move for each scenario. Once trained, the ANN can be used by the chess engine to evaluate quickly and efficiently the score for each possible move — ANN-based chess engines become even more positional. And boring.

Google’s AlphaZero changed the landscape. It implements a new ANN-based approach that can be trained to play not just chess, but other board games. Given a game state, the AlphaZero engine computes a policy that maps the game state to the probability distribution of making each of the possible moves. In chess, that’s 4,672 possible moves — for white’s first move. For human players, this is ridiculous: There are 20 legal moves for white’s first move. And that’s the point.

DESIGN & PROMPTS: : Anas Albounni, KUST Review IMAGES: AI Generated, KUST Review.

AlphaZero includes all sorts of moves that are illegal, like selecting empty squares, selecting opponent’s pieces, making knight moves for rooks, or making long diagonal moves for pawns. It also includes moves that pass through other blocking pieces.

During training, nothing is learned or imposed about avoiding non-valid moves. The engine just post-processes the ANN output, filtering out illegal or impossible moves by setting their effective probability to zero. Then it re-normalizes the probabilities across the remaining valid moves. A philosopher could argue that this engine is devoid of ethics as it does not distinguish illegal from impossible, but the resulting ANN structure is simpler than one expressing only valid moves. On a modern processer, AlphaZero needs just a few tens of milliseconds to make a move.

This speed enabled AlphaZero to play against itself in millions of games, completing its training with reinforcement learning, which privileges moves that lie on a sequence that led to victory in the past. Of course, to know whether a move is on a winning sequence, one must complete the game, so AlphaZero reinforcement was performed by playing “fast and dumb,” i.e., using a very shallow search depth. Playing dumb in the reinforcement training phase maximizes the number of games that end in victories and defeats rather than in uninformative draws. As a result of this, AlphaZero considers fewer positions than the algorithmic chess engines of the past.


A classic chess engine may well consider tens of millions of alternative moves per second, but its playing style is boring.

AlphaZero and successors like Deep Chess have a distinctive style, steering away from merely seeking positional or material advantage. Their style is alien — best likened to atonal music: difficult to appreciate for anyone but the chess elite, and certainly useless for the average human amateur to learn or improve their game.

It is interesting that we humans still describe an intelligent chess engine’s style in positional terms.

Like the ‘90s Deep Blue victory, today’s post-AlphaZero scenario highlights some general problems that we will have to solve to be able to work together with the super-human AI engines of the future. AI decision-making must be intelligible to humans for us to accept its decisions. This interpretability needs to be wired into the AI training. Plus, interacting with humans is a crucial step for AI engine evolution: Playing against all possible competitors makes them stronger than any human can individually hope to become.

The game of chess, always a metaphor for life, suggests that controlling the evolution of future AI engines may become more akin to taming a tiger than training a pet.

Ernesto Damiani is senior director of the Robotics and Intelligent Systems Institute at Khalifa University.

DATA to delivery

Welcome to Industry 4.0, considered by many experts to be the fourth industrial revolution. Artificial intelligence and data analytics are a big part of it and are already changing how supply chains work. Here are just some of the ways they make getting a product from the manufacturer to your home cheaper and more efficient.

IN THE FACTORY

Generative design: An algorithm receives design parameters (such as cost and information on available materials) and generates thousands of options to find the best one.

LISTEN TO THE DEEP DIVE

Order management: AIs handle complicated order information from multiple channels.

Quality control: Sensors inspect products for defects.

Predictive maintenance: AI monitors systems and machines for early signs something is about to break down, preventing expensive factory shutdowns.

Compliance management: AI manages the red tape when the same product is sold in different markets with different regulations.

Customization: AI may be used to create such customized orders as bespoke suits and made-to-order shoes. And in a process called “reshoring” or “nearshoring,” products made far away can be customized closer to the sale point at the last minute.

IN THE WAREHOUSE

Stocking: Digital cameras monitor inventory levels and AI robots pick, sort and pack products.

Finding damaged packages: Machine learning models scan and analyze images to spot damaged objects.

Helping workers with wearable technology: Smart glasses “read” barcodes. Natural language processing helps humans work hands-free to pick items more safely.

THROUGHOUT THE PROCESS

Supply chain visibility: Internet of Things (IoT) devices provide instant information about such conditions as the location and temperature of shipments. Businesses can spot bottlenecks, manage disruptions in real time and make data-driven decisions.

Collaborative supply chains: Multiple companies use data and analytics to work together to plan and execute supply chain operations. The cooperative approach allows the companies to serve similar customers or achieve a common goal.

DELIVERIES

Optimal routes: Vehicle routing algorithms (without problems) use such factors as capacity, delivery priorities and time windows to plot the most efficient routes.

Real-time conditions: AI can monitor weather, traffic and other conditions to reroute as necessary.

Autonomous vehicles: Truck platooning technology can permit a group of vehicles to operate extremely closely, reducing wind resistance and decreasing fuel consumption for transportation between factory and warehouse or retailer. Smaller vehicles will be used for deliveries. Algorithms optimize routes while AI helps vehicles avoid collisions.