Future of Humanity: AI & Robotics | Free Documentary

Summary

In this fascinating documentary, the future of AI and robotics is examined, revealing a world where robots are no longer confined to industrial settings but are now integral to various aspects of life, including homes, education, and even medical diagnostics. The documentary highlights advancements in humanoid robots that interact socially, bio-inspired robotic animals, and autonomous vehicles poised to revolutionize transportation. Ethical dilemmas and the limits of AI, such as decision-making without human oversight, are also explored, prompting viewers to consider the implications of a future dominated by artificial intelligence.

Highlights

• Robots are now in homes, schools, and hospitals, acting as companions and helpers 🤝.
• Humanoid robots like Pepper and Nadine are designed to recognize emotions and interact socially 😊.
• Bio-inspired robots are mimicking real animal movements, offering incredible advancements in robotics 🐦.
• Autonomous vehicles are designed to reduce road accidents and traffic congestion, hinting at a safer future 🚦.
• AI's rapid progress in processing data is transforming industries, especially in surveillance and healthcare 📊.
• Deep learning enables machines to learn and adapt, moving AI closer to human-like problem-solving 🧠.
• Ethical dilemmas arise as robots decide without human input, urging discussion on AI's future role in society 📜.
• Despite advancements, AI struggles with human-like creativity and emotional intelligence, areas where humans excel 🎭.

Key Takeaways

• Robots are stepping out of factories and into everyday life, even acting as social companions and educational aides 🤖.
• Bio-inspired robots mimic animals to perform tasks in natural environments, showcasing nature-inspired innovation 🐟.
• Autonomous vehicles are advancing rapidly, promising a future with fewer road accidents and traffic jams 🚗.
• AI has surpassed human abilities in data processing, enhancing fields like surveillance and medical diagnostics 🏥.
• Ethical concerns about AI decision-making without human oversight highlight the need for careful regulation ⚖️.
• While AI shows remarkable progress, it has yet to replicate human creativity, emotion, and complex decision-making 🎨.

Overview

The documentary delves into the transformative role of AI and robotics in modern society. It explores how robots, once confined to repetitive tasks in factories, are now increasingly part of our daily lives, with roles in households, schools, and healthcare. This shift is marked by the emergence of humanoid robots that can engage socially, offering companionship and performing tasks previously heralded by humans.

Bio-inspired robots take cues from nature, replicating animal forms and behaviors to meet unique challenges in diverse environments. From robotic fish that seamlessly integrate with marine life to aerial robots like the robotic bat, these machines highlight innovative strides in robotics. Autonomous vehicles are also introduced, promising a reduction in human error and traffic mishaps, moving towards a future of efficient, accident-free travel.

The documentary also raises significant ethical questions. As robots become more intelligent and autonomous, concerns about decision-making without human intervention arise. While AI excels in data handling and pattern recognition, it lacks the intricate depth of human emotions and creativity. The film invites viewers to ponder the future role of AI, emphasizing the need for conscious choices about technology's path and its integration into society.

Chapters

• 00:00 - 01:00: Introduction to Robotics and AI The chapter titled 'Introduction to Robotics and AI' explores the fascinating world of robots and artificial intelligence. It draws a parallel to sci-fi movies, highlighting the incredible advancements being made in laboratories worldwide. The chapter notes that future robotic designs will mimic humans, animals, and other yet undefined forms, equipped with unexpected capabilities.
• 01:00 - 02:00: The Role and Capabilities of Robots The chapter explores the role and capabilities of robots, particularly focusing on their form of intelligence. Robots can learn and improve by recognizing underperformance and adapting to do tasks better through practice. The chapter questions the secrets behind these transformative technologies and considers their impact on human life. It suggests a future where human roles are minimized to providing expertise, with automation handling the rest, highlighting the current and potential capabilities of robots.
• 02:00 - 03:00: Social Robots: Integration into Daily Life The chapter titled 'Social Robots: Integration into Daily Life' introduces the concept of a future where robots and artificial intelligence are deeply integrated into human life. It discusses the advancements and capabilities of robots, including AI and big data, which enable them to perform logical operations, make deductions, and predictions. The narrative paints a picture of a world where robots are an integral part of society, working alongside humans, and the line between human and machine becomes increasingly blurred.
• 03:00 - 06:00: Humanoid and Pet-like Companion Robots The chapter delves into the advent of companion robots that are humanoid or pet-like, exploring how they are increasingly becoming a part of our daily lives. The discussion addresses the readiness for this technological evolution and how music and entertainment integrate with these robots. This transition is portrayed as a future that is not just impending but has already begun, urging readers to consider their preparedness for a world where such robotic companions play a significant role.
• 06:00 - 09:00: Biologically-Inspired Robots The chapter titled 'Biologically-Inspired Robots' begins with an enthusiastic reception, indicated by applause. This sets the stage for an engaging exploration into the world of robotics that draw inspiration from biological systems. The transcript likely delves into various aspects of how living organisms and their behaviors inspire the design and function of robots. Topics could include biomimicry, the advantages of biologically-inspired designs over traditional robotics, and potential applications in various fields. The chapter possibly covers specific examples of biologically-inspired robots, discussing their features, how they mimic nature, and the problems they aim to solve. It might also explore future trends in robotics and ongoing research in this fascinating intersection of biology and technology.
• 09:00 - 14:00: Autonomous Vehicles and Transport Systems The chapter "Autonomous Vehicles and Transport Systems" explores the evolution and integration of robots beyond industrial applications. In the 21st century, robots have expanded their roles from performing repetitive factory tasks to becoming integral parts of daily life, impacting streets, schools, homes, and entertainment venues. This expansion signifies a shift in how society views and utilizes robotic technology, transforming these machines from basic tools into complex entities that influence various aspects of human life.
• 14:00 - 18:00: Challenges and Ethical Concerns in Robotics The chapter explores the integration of social robots into daily life, emphasizing their role in collaboration and communication with humans. Despite their playful appearance, these automatons are more than just novelties, as they are expected to become active participants in various aspects of human life.
• 18:00 - 21:00: AI in Surveillance and Medical Diagnosis The chapter discusses the deployment of a device named Temi, which has been launched in the United States. The device is equipped with advanced technologies including cameras, lidar for mapping environments and obstacle avoidance, a screen, and an intelligent communication system. This system enables Temi to recognize individuals, respond to calls, and exercise control in its operations.
• 21:00 - 25:00: The Future of AI and Robotics: Opportunities and Limitations The chapter explores the potential and challenges associated with the integration of AI and robotics into daily life. It highlights the emergence of AI-powered domestic assistants that are capable of connecting with smart home devices and even contacting loved ones through voice commands. These technologies come with a substantial price tag, showing that they may become a commonplace aspect of future homes. Additionally, the chapter touches upon the advancements in robotics on a global scale, specifically mentioning a groundbreaking development in Saudi Arabia, where researchers have successfully created the first robot that can converse and comprehend Arabic, marking a significant milestone in making robots more culturally and linguistically adaptable.

Future of Humanity: AI & Robotics | Free Documentary Transcription

• 00:00 - 00:30 they're like something straight out of a sci-fi movie yet these creatures are real come to life in laboratories all over the world the shape of things to come will resemble humans animals and some things we've never seen before equipped with surprising
• 00:30 - 01:00 capabilities including a form of intelligence by doing a task the system noticing that it's not performed as well as it could it can learn and change so that it does it better the next time they adapt through practice what are the secrets behind these technologies that are changing the world and how will they transform our lives humans are going to be slowly reduced to providing human expertise everything else is going to be automated robots can do so much for us already and
• 01:00 - 01:30 they will do so much more in the future welcome to a new world where robots are among us where artificial intelligence is our new ally where vehicles have no human driver and where the border between man and machine slowly disappears ai and big data allows this kind of robot to be very logic and to make deductions and predictions this world is our reality
• 01:30 - 02:00 and the future has already begun are you ready [Music] [Music]
• 02:00 - 02:30 [Applause]
• 02:30 - 03:00 in the 21st centuries robots have left the factories and the labs they are no longer confined to simple repetitive tasks today these creatures of metal and plastic have permeated almost all areas of our lives you can find them in the streets in schools in the heart of our homes and even in concert halls no longer considered simple tools modern
• 03:00 - 03:30 machines can collaborate with us and even communicate with us we have entered the era of social robot despite their playful appearance these little automatons are not just novelties in the near future they will participate with us in daily life [Applause] the most recent iteration has premiered
• 03:30 - 04:00 in the united states hey tammy hello john it's called temi and was launched onto the market with a considerable publicity campaign equipped with cameras a lidar to map its environment and avoid obstacles a screen and an intelligent communication system it can recognize you come when you call it and can control
• 04:00 - 04:30 connected devices in your house and it can contact your loved ones with a simple voice command at two thousand dollars a piece it just may be your next domestic assistant but the united states is not the only place where our new companions are kept busy on the other side of the globe in saudi arabia researchers have created the first robot capable of having a conversation in arabic and understanding
• 04:30 - 05:00 local dialects how are you i am fine my first time at the saudi electronic university in saudi arabia the goal to use it as a receptionist in hotels or shops and even as a teacher the use of a robot with students has already been tested in classrooms in finland
• 05:00 - 05:30 elias a small humanoid robot 60 centimeters tall can walk talk dance and recognize the faces of children and their emotional states are you sad no i'm not oh i was wrong how are you feeling today then i'm happy that's great remember the words
• 05:30 - 06:00 if a child is bored elias can offer fun activities like singing or dancing ice cream he provides relaxation exercises if a child is tense or irritated kids appreciate this little teaching assistant who engages them through play
• 06:00 - 06:30 and repetition robots at school the idea seems completely improbable and yet the results with students are promising robotics specialists are convinced these experiences will become a bigger part of our future my name is daniela russ i'm a roboticist the director of the computer science and artificial intelligence laboratory at mit and a professor at mit
• 06:30 - 07:00 there are activities in the classroom that require routine uh interactions and those activities could be very well delivered with machines in fact i believe that robots and studying robotics is such an important aspect of 21st century education for a variety of reasons first because i think that robots are magical to children and they just get excited about
• 07:00 - 07:30 thinking about robots and they really um also i believe like the idea of being able to control something when in fact they're being controlled all the time by the adults in their environments elias the small resident of the finnish school was designed by softbank robotics one of the largest robotics companies in the world its engineers have already designed many
• 07:30 - 08:00 robots in their laboratories but their new star is this one pepper bonjour pepper is one of the first robots able to identify both faces and emotions and adapt its response to the person it communicates with today the robot is already used to
• 08:00 - 08:30 assist customers in shops train stations and museums it is being tested in hospitals and retirement homes to help guide patients [Music] pepper is actually one of the first representatives of a new generation of companion robots small automated beings specifically developed to form relationships with humans i am alexander mazel director of innovations at softbank robotics [Music]
• 08:30 - 09:00 what interests us most is moving towards long-term companions in one of our scenarios when i speak to the robot i say hello the next day when i see him something will change in our contact he will say hello but it's not the same hello as today that is to say that little by little he adapted he changed a little of his behavior according to what i told him he's going to give me winks and things like that tap my sight
• 09:00 - 09:30 i really believe in the robot pepper as a companion for the elderly because we know we have this need to help people who crave conversation human contact i would say people with deep loneliness we still have lots of things that are feasible like adapting a little to the rhythm of a person's life that is to say the robot goes to bed at the same time as the elderly person you can also memorize the usual bedtime and offer a gentle reminder say you should go to bed now usually you go to bed earlier there are a lot of little things like that that
• 09:30 - 10:00 really make a difference knowing there is someone in the room who can be of service who can decide to bring a drink render small services the human-to-robot relationship could possibly resemble a human-to-human relationship so much so that some scientists wish to go even further by developing machines with a nearly human appearance hello
• 10:00 - 10:30 hello how are you i am a social companion i can speak with emotions and i can recognize people this is nadine the latest generation in social robots capable of recognizing its conversation partners remembering past discussions and analyzing emotions and starting a conversation how many languages can you speak i can speak in hindi english chinese german french and
• 10:30 - 11:00 japanese a robot made in the image of its designer nadia thalman i am nadia talman i am a roboticist in mira lab at university of geneva nadine capture information from a face and then this information goes to a big data set where it is detected which kind of emotion it is then it is analyzed through a model of emotions and then nadine behaves
• 11:00 - 11:30 according to what she has understood for example if i say i'm sad i have very bad news she can analyze the speech and the intonation but she can also analyze my emotion and when she will answer she will take a tone that is with compassion and also what she will say is something that is also relatively neutral or sad do you have emotions i do not feel emotions but i am
• 11:30 - 12:00 programmed to simulate the full range of human emotions i can be happy with a smile you know myself i am entering soon my third age and i really like to have a person to take care of me i know that is not possible 24 hours so in this case particular case i'm happy that a humanoid is able like a human to deal with me in a natural way
• 12:00 - 12:30 nadine can you tell me a joke why shouldn't you write with a broken pencil because it's pointless thank you for the joke and thank you for the conversation like pepper and even more like nadine humanoid-shaped robots occupy a special place in the field of companion robots according to certain experts a human-like form is one of the keys to fully integrating companion robots into our daily lives
• 12:30 - 13:00 it's very important to have humanoid shape like humans because when we have an interaction with someone is always with a human so i never speak to a box i never speak as such as a cat and i don't expect the intelligence to we could put similar software with a box and a pet but myself i'm absolutely not motivated to live with a box or a pet that is intelligent because it doesn't correspond to
• 13:00 - 13:30 to the natural interaction i have with humans today people ask us why is your robot better than a box with speakers and for us we try to emphasize the physical aspect the fact that pepper has a body when a child is crying you immediately want to take them into your arms what interests us most is long-term companions and that's what we want to explore in robotics but this debate has divided the robotics community some experts believe that designing
• 13:30 - 14:00 machines too close to our own image could provoke a feeling of rejection my name is alkai spirit i'm a bioroboticist and i'm leading a biorobotics lab at epfl in lausanne switzerland it should not be too lifelike because then people find it creepy it's called the uncanny valley so maybe it's a bit more better than the face is a bit cartoon-like than too realistic but indeed having eyes to look at having a
• 14:00 - 14:30 mouse that moves that that probably is is more engaging for interaction so for that clearly having a humanoid phase a bit more cartoon than realistic i think can be very good alternatively it could be a pet-like robot maybe a little dog-like robot or something that you feel is cute and and he's still there to assist you in some way companion robots in the form of pets the idea has already occurred to some designers
• 14:30 - 15:00 in the south of england the templeman house retirement home has recently adopted a new artificial resident [Music] his name is biscuit and he behaves like any little dog and can interact with residents at any time of the day to provide comfort [Music] yes you are a beautiful one
• 15:00 - 15:30 [Music] on the other side of the atlantic in florida alzheimer's patients also connect with robotic animal companions studies have shown that it helps to reduce patient anxiety as well as the use of psychotropic medicines and painkillers beyond these adorable robotic stuffed animals the real animal world provides
• 15:30 - 16:00 an unlimited source of ideas [Music] and with the development of bio-inspired robots we are now witnessing machines capable of faithfully reproducing animal movements and the ability to interact with their environment the animal kingdom is so beautiful you have so many morphologies for adapted for different environments it's really worth exploring all these different morphologies
• 16:00 - 16:30 nature inspires and challenges ours nature has achieved so much in terms of movement in terms of reasoning capabilities our machines are very simple as compared to what nature achieves yet we aspire to get as close to nature as
• 16:30 - 17:00 we can [Music] [Music]
• 17:00 - 17:30 today scientists are able to reproduce the way animals move on land under water and in the air [Music] [Music] in germany festo a company specializing in automation has developed a flying
• 17:30 - 18:00 robot that mimics the aerial capabilities of a bat in every respect its wings are made of a thin elastic membrane and can be controlled separately by two small independent motors with a wingspan of 2.28 meters and a weight of only 580 grams this bat can navigate with the same dexterity as a real bat
• 18:00 - 18:30 [Music] for the time being it still needs human assistance to take off and land but it can fly autonomously in a defined space thanks to its cameras and numerous on-board sensors for now these artificial creatures are mostly confined to laboratories only a few prototypes have been deployed in the field but they have already demonstrated their usefulness
• 18:30 - 19:00 [Music] [Music] in the netherlands the row bird a robot
• 19:00 - 19:30 that mimics the flight of the peregrine falcon was created it is already being used in several airports around the world to avoid collisions by scaring birds away from aircraft [Music] a new species of robotic animal has been born [Music] sophie is a soft robot
• 19:30 - 20:00 with its silicone tail two motorized fins and a propulsion system based on pumping and ejecting water it can move and change speed and direction very smoothly almost like a real fish [Music] i wanted a robot that could not just sit there transmitting images of underwater ecosystems but a robot that could join an underwater ecosystem a robot that
• 20:00 - 20:30 could join a school of fish a robot that could really help us speak into the secret lives of the underworld and so we created sophie where sophie stands for soft fish and this is a robot fish whose tail is soft controlled remotely by a diver located approximately 20 meters away sophie can explore the heart of the coral reefs at close range
• 20:30 - 21:00 photographing the biodiversity without disturbing it thanks to the camera hidden in its nose this fish has been an extraordinary experience because indeed we have been able to take the fish to the ocean and our fish was able to swim side by side with real fish uh without disturbing them and the fish was able to collect uh really special images about the underwater world
• 21:00 - 21:30 i really believe that if we're going to have robots working side by side with people or working side by side in natural environments those robots have to have bodies that fit with those environments control systems that fit with those environments and also modes of interaction that are intuitive and fit very well
• 21:30 - 22:00 we still have a long way to go before we can produce the robots that integrate perfectly into their environment and help biologists [Music] nevertheless like sophie some prototypes are already becoming valuable tools of science
• 22:00 - 22:30 aurobot a replica of a giant 300 million year old lizard the orobadus pubsti scientists used the skeleton and fossilized footprints of the animal to bring this extinct species back to life here what we did is make a robot that exactly mimicked and replicated that morphology and we wanted to answer this scientific
• 22:30 - 23:00 question of what was the most likely gate way of moving of that robot especially since we knew the footprints following in the footsteps of its distant ancestor aurobot allowed researchers to discover the locomotion of this prehistoric vertebrate and interestingly it was a fairly agile gate fairly athletic more closer to the cayman than to the salamander for instance so much more modern than we were thinking in buying spy robotics people always
• 23:00 - 23:30 think you you you need to take in spray from biology and and contribute to robotics but i think there's a wonderful feedback loop where the robot becomes a very nice tool for biology in general so i think a robot can be an interesting tool for studying biomechanics how the body interacts with the environment for paleontology how previous animals were locomoting or for neuroscience as a tool to test hypothesis for instance how the brain interacts with the environment and i really love this concept of having a robot as a tool because of course many
• 23:30 - 24:00 people use numerical simulations do it in simulation but once you have a robot and start interact exploring the interaction of all the components you really have a wonderful tool to very systematically and quantitatively test hypothesis [Music] and still robots continue to improve moving ever more gracefully
• 24:00 - 24:30 [Music] exploring their environment [Music] and even using tools [Music] in many ways machines are gradually learning to replace humans in specific activities
• 24:30 - 25:00 some researchers are hoping to go even further their goal to give these robots the ability to make their own decisions [Music] moving over challenging terrain is the primary job of these autonomous robots
• 25:00 - 25:30 the centaur this half-humanoid half-quadruped robot inspired by the mythical creature is one of the most recent examples of a rescue robot it can grab and move objects with its two arms and navigate rough terrain on four legs [Music] a fully autonomous version is still in the works but it will soon be tested under real-life conditions in damaged
• 25:30 - 26:00 buildings or factories places too dangerous for human rescuers a decision-making ability isn't just for humanoid robots or even for bio-inspired ones the next wave of thinking robots might just turn up in your garage
• 26:00 - 26:30 new generations of autonomous cars are loaded with sensors video cameras detect traffic lights read signs and even distinguish between pedestrians and bicycles radar and lasers pinpoint surrounding objects lidar a 3d mapping radar collects data which is analyzed by onboard computers [Music]
• 26:30 - 27:00 those computers will use increasingly sophisticated artificial intelligence [Music] clunky awkward prototype cars that run on test tracks are a thing of the past driverless vehicles are about to merge into mainstream traffic to the surprise of some experts
• 27:00 - 27:30 [Music] i'm paul rosenblum a professor of computer science at the university of southern california i was shocked at how quickly we got from where it was almost impossible to get a computer to do a reasonable job driving a car to the point where we were talking about having them out on the streets that was something that just went enormously fast
• 27:30 - 28:00 developments in the field of autonomous cars suggests we're on the road to a safer future where machines would eliminate human error in this connected world everything is designed to reduce the risk of road accidents to nearly zero it will become safer because we will have robot cars connected with other robot cars and with human driven cars we may
• 28:00 - 28:30 have sensors installed on buildings on the infrastructure so that even the camera on the car cannot see what is happening around the corner the car can talk to the camera installed at the corner and see that a child is about to run in the middle of an intersection and adjust to that [Music] just imagine how wonderful will be to have a world with no road accidents because our cars are so
• 28:30 - 29:00 aware of what is happening on the road that they will always avoid a road accident and if you want to ride and drive your car then maybe the autonomy system will be sort of like a um like a parallel autonomy system that will look over your shoulders like a guardian angel and prevent you from making mistakes even if you're a little bit distracted [Music]
• 29:00 - 29:30 a world without accidents but also a world without traffic jams because autonomous cars also offer large-scale solutions something impossible to imagine with human drivers with autonomous vehicles we hope we will manage uh congestion in ways that will ensure we're never stuck in traffic now how to do that well i think that requires a much broader view of what is
• 29:30 - 30:00 happening in the transportation system there are new types of algorithms that ensure we can control who goes where and when these are called fleet management algorithms that will relieve a great traffic burden from our road networks the ultra-dynamic field of autonomous vehicles doesn't stop at cars
• 30:00 - 30:30 around the world other means of transportation are currently being tested in germany after two years of research engineers have managed to create the first 100 autonomous motorcycle it can start accelerate drive in straight lines make turns and maintain
• 30:30 - 31:00 its balance without any human intervention [Music] in sweden a startup has just launched teapod a driverless truck seven meters long its cabin which houses the electric motor doesn't even need a windshield
• 31:00 - 31:30 it can transport up to 20 tons of cargo drive independently or be controlled remotely during more complex journeys recently its little brother the t-log made its public debut
• 31:30 - 32:00 specially designed for transporting wood its cabin has been removed to facilitate very large tree trunks vehicles like teapot and t-log may revolutionize the heavy goods vehicle sector though the development of these autonomous trucks isn't limited to trucking
• 32:00 - 32:30 at this norwegian airport fleets of unmanned snowplows clear the runways operated in real time from the control tower machines 20 meters long and 5.5
• 32:30 - 33:00 meters wide can clean 350 000 square meters in just one hour [Music] and it isn't just autonomous vehicles that are being developed in the netherlands the first autonomous
• 33:00 - 33:30 boat prototypes are in the testing phase the goal to transport passengers or goods on the amsterdam canals with their modular design they can join to form floating bridges or platforms
• 33:30 - 34:00 but while all these autonomous modes of transport could soon become a regular site on our roads and rivers the question remains can we trust them with our lives [Music] [Music] a lot of the hard part in autonomous driving is is the sensory processing how do you interpret what's going on
• 34:00 - 34:30 around you and if you misinterpret it you can make fatal mistakes so a lot of the trouble there really is what kind of sensors are using how accurate an image you can get of the world around you from those sensors and therefore are you making appropriate decisions given the true situation not just the situation you think you're in so you have to decide for example do i want to go left or right if i'm trying to go somewhere so there's route planning which is part of the big problem of intelligence
• 34:30 - 35:00 there are the hard ethical issues if you're in a dire situation and you have to decide whether to crash your car or run someone over their decisions about how cautious you're going to drive um if the car is behind you a hard kick because you're going too slow and being too cautious should you speed up so when people talk about true autonomous driving you've got to deal with all of that [Music] these questions are all the more crucial today as vehicles and other autonomous
• 35:00 - 35:30 robots enter a new era deep learning expressiveness so emotions head position if your head is going down hi i'm rafael d'andrea i'm ceo and founder of verity studios and i'm a professor at eth zurich by doing a task the system noticing that it's not performed as well as it could it can learn and change so that it does it better the next time whether the learning takes place on the vehicle itself whether that information is
• 35:30 - 36:00 transferred to a computing station whether it's to the cloud and all of this information from any vehicles is shared to improve it's really the system learns and adapts deep learning is rightfully the star of ai right now they're generating huge successes um based on the availability of lots of data and lots of computing and and some improved algorithms with deep learning algorithms can process and classify data independently
• 36:00 - 36:30 they can learn and progress through experimentation just like human beings the concept gets tested with this experiment on a small country road in england where a car will learn to drive on its own at the start of the test like a driving school teacher a human driver straightens the wheels when the car deviates from its path
• 36:30 - 37:00 and the result speaks volumes it took the teacher only 20 interventions before the car learned to follow the road we can drive the car equipped with three cameras on the roof and an on-board system had not been previously programmed for
• 37:00 - 37:30 autonomous driving it learned to drive as any beginner would do [Music] in the united states a robotic hand undergoes an unprecedented exercise it's instructed to rotate a cube until a certain letter shows on the top
• 37:30 - 38:00 as the experiment progressed the hand accomplished its task faster and faster [Music] improving its skill just like we humans learn to do by repeating action [Music]
• 38:00 - 38:30 animal is a robot designed to inspect industrial sites as seen in this test performed on an oil platform inspired by how a dog moves it can walk run climb stairs and keep its balance or stand up in any circumstances thanks
• 38:30 - 39:00 to its fully rotating body equipped with 3d scanners it analyzes its environment can use its legs to interact with objects and above all learns from its errors over time anymore becomes faster and more efficient [Music] today's machines can solve increasingly complex problems but is there a limit will they ever be able to make appropriate decisions in changing
• 39:00 - 39:30 situations [Music] can artificial intelligence ever develop common sense [Music] or resolve moral dilemmas for experts we may have reached the limits of what a machine can be trusted to do [Music] the big challenge in all these situations is generalization how can you
• 39:30 - 40:00 extrapolate what you have learned so far to new conditions that you have never seen and there having a human supervisor is still i think by law needed and because that that's you can a program can make make mistakes in extrapolation and and it's called this problem of of overfitting you learn too much on the sub problem and then that gives very bad solution for other problems so this notion of generalization to all possible conditions is is really really tricky as far as i know you both have to
• 40:00 - 40:30 understand your learning algorithms understand the biases in your data but you also have to add some protections what we in humans would cause comments would call common sense so you don't just learn anything and you say oh that's what i learned in this situation so i would go with it you bring your other knowledge to bear and say is that appropriate is that reasonable will something go wrong if i do it that way even though that's sort of what this situation taught me if we don't have learning computers do
• 40:30 - 41:00 that then they're going to do some seriously bad things the ethics of decision making are at the heart of artificial intelligence but at the same time the range of decisions made by machines continues to expand in australia an underwater robot has received authorization
• 41:00 - 41:30 to hunt and kill rangerbot's mission is to identify a particular species of invasive coral eating starfish and to inject it with a deadly poison after two years of development it carried out its first field tests in the heart of the coral reef the results are promising the robot detects the starfish with more
• 41:30 - 42:00 than 99 accuracy it could help to save the endangered australian reef [Music] but the ranger bot also poses an important ethical question can a robot decide to kill a living being [Music] this notion of robots acting on the environment and possibly indeed the
• 42:00 - 42:30 destroying part of the environment that we we feel is is is dangerous is is indeed tricky um and the starfish example is i think very interesting i think it's very good to protect the the coral reef and and that and indeed in this case i would suggest to have a human in the loop somehow to not let the system completely independent to to really avoid big mistakes and start destroying things that we should not do i would always suggest to have a human in the loop somehow a human supervisor for all all things where where the robot is supposed to do
• 42:30 - 43:00 something a bit dangerous during the environment likely what would happen is that you know let's say that this use case was successful with a person monitoring it and after several years you know you try it autonomously and then you you know a person is supervising it and then after enough time people are comfortable that it works then once you once you're there then maybe you can move to that level but frankly you know why would you ever want to remove a person from that situation you know is
• 43:00 - 43:30 it is it really that costly to have a person supervising that process i think just from an ethical and um you know moral perspective i think it's always good to include people in the loop whenever you have something that is affecting life whether it's human life or animal life morality and ethics are the domain of humans not robots machines cannot be held responsible for
• 43:30 - 44:00 their actions it is up to us to set the limits to avoid lapses beyond these legitimate fears looms a larger question while robots and artificial intelligence systems are becoming more efficient will they eventually overtake us ai is already proving to be much better than humans in certain areas especially in data processing
• 44:00 - 44:30 machines can recognize and categorize information faster and more efficiently than any human brain they're a boon to industry and scientists [Music] and one of the fields that benefits most from artificial intelligence is of course surveillance
• 44:30 - 45:00 [Music] like this software capable of detecting a revolver on video surveillance images in a fraction of a second and alerting authorities [Music]
• 45:00 - 45:30 [Applause] or like this system developed by a british startup to counter cyber attacks mimicking the human immune system it gradually learns to identify and defend against threats from hackers
• 45:30 - 46:00 it has already proven itself in british ports and in some power plants another area where artificial intelligence may be a game changer is in medical diagnosis [Music]
• 46:00 - 46:30 researchers at the university of oxford have developed ultronics software capable of identifying markers of heart disease on scanners it can make a clear and reliable diagnosis in seconds much faster than doctors will robots be the doctors and police of tomorrow does progress mean creating artificial
• 46:30 - 47:00 beings more intelligent than ourselves do you use ai i am using ai all the time to understand what you say to generate answers to recognize faces and emotions for example wow the answer is not unanimous i don't know that we can build things that are a lot more intelligent than people in general we know we can build specialized systems that can outperform people uh whether it's now in games like chess and go or whether it's in other kinds of
• 47:00 - 47:30 things what we don't know is that it's possible to build general intelligence that is just much more powerful than people and if i were betting i would say yeah maybe a bit above i mean we can certainly do as best as the smartest human uh how much smarter the machine can get i really don't know at this point i think robots will never be as good at humans in anything that requires a lot of imagination art clever thinking about design or
• 47:30 - 48:00 construction of things there we have something special i think we will keep uh for a long time of really being creative and then i would say it will still take many years to get like a robot being able to play tennis like roger federer or playing football or climbing the everest like a human humans and animals in general have locomotion skills that are so amazing so it will take very very many years to be able to replicate those so i think there are some aspects where
• 48:00 - 48:30 as humans we will still do better than robots for a long time [Music] today it's true that we often talk about our fear of robots that we'll have super ai and be much more intelligent than humans and will take over but in practice the most important use of robots is for repetitive simple tasks like robots that pick up tree leaves from the ground when we talk about education they can repeat multiplication tables for hours and hours to children and these tedious human tasks will have robots doing them
• 48:30 - 49:00 [Music] the question of the role that robots and artificial intelligence will play in our lives is not new but the technological context is changing making the issues and the stakes more and more crucial ultimately it is up to us humans to choose carefully what we want to do with the machines we create
• 49:00 - 49:30 how far do we want to take them and where will they take us it is important to remember that robots ai systems machine learning systems are tools they are tools by the people and for the people they are extraordinary tools that can do so much but like any other tools they are not inherently good or bad they are what we choose to do them and
• 49:30 - 50:00 we can choose to do extraordinary things but as we think about technology evolving it is also very important to think about the consequences of our work and i don't believe we can stop technology from expanding but what we can do is to stop and think about the consequences and put in place the rules and regulations that ensure that the use is for the greater good [Music]
• 50:00 - 50:30 you