2028 u. S. Election. Only just the. How is Artificial Intelligence shaping our society from offices to living rooms devices with human like characteristics are quickly creeping into our lives the besides the benefits what are the risks and will they i take over the world one day this is inside story. Hello and welcome to the program im peter w. Now many of us interact every day with siri apples voice activated digital personal assistant siri can find information gives us directions sends messages plays music one example of how ai is becoming a bigger part of everyday life chances are its on all your devices can mimic areas of Human Behavior and could soon learn about our feelings and our emotions the key sentence there the key word mimic the machines are getting smarter of course but experts are divided as to when well move from basic ai to the scary stuff of Science Fiction if at all but there are real and relatively immediate threats almost 50 percent of american jobs are under threat over the coming 15 years and health care will probably get more expensive as ai increasingly gets involved in keeping you alive. Well banks are using it to detect fraud and predict changes in the stock markets is used by air Traffic Controllers to help ensure Flight Safety in the air and on the ground its helping the Police Forces around the world identify suspects from c. C. T. V. Images ai is also helping self drive cars to navigate road system one example of course tesla with its predictive capabilities machines that can identify images of being used by doctors to spot disease jobs done by humans are under threat to and they are increasingly being used to decide whether you will even get an interview in the 1st place so is the future excel or its towards us may be worth reminding ourselves the great design and great art well thats always been the reserve of us human beings biology it wasnt a machine that gave us the Sistine Chapel the pyramids of egypt or the works of kabbani aljazeera as Gabriel Elizondo has been to the designs for a different future exhibition it aims to encourage people to imagine a coming way of life very different to what we have today its an Art Exhibition not about the past not about the present but about imagining the future of everything like this 25 collaged passports proposing a system to allow people to temporarily exchange citizenship a critique of a world where goods and services freely cross borders where humans often cant. Its one of more than 75 exhibits by designers tackling issues of the future and the human condition at a major new exhibition at the pill adelphia museum of art it is a vast show that tries to ask a lot of questions without necessarily giving answers. When you come to an exhibition about the future you might expect to see things like this robots and there is one this is corey that looks at the interaction between robots and humans and Artificial Intelligence but this exhibition is also about a lot more these faces were created from d. N. A. Extracted from discarded cigarette ends and gum on the streets the designer using the Genetic Information to render the portraits there are several pieces related to food like these 2 salmon a larger genetically modified one next to a smaller organic one and a replica of stakes grown out of human cells from hospital waste both provocative ideas be flexing a future where food resources are strained the idea is to make people think about food as a cornerstone of human civilization and think about what will happen to food and the future of human diet as we start thinking about the big challenges that are lying ahead like Climate Change diversity loss and environmental pollution einstein once said he never thinks of the future because it will come past enough in this exhibit designers are the ones bringing bare ideas for the future and not telling people what to think rather what to think about gabriels on doe. Philadelphia. Ok lets bring in our guests joining us from mass tricked on skype is catalina go on to assistant professor in private law at mass tricked university she also could manage is the master of law and tech lab at the boys ray is associate professor in residence at the Northwestern University here in cattle and author of automated journalism algorithms bots and complete ational called the titian and in berlin on skype caroline cinders is Machine Learning designer and a fellow at the Missoula Foundation and Harvard KennedySchool Welcome to you all caroline can i come to you 1st what we have at the moment is euphemistically known as narrow a what does that mean and perhaps i may not be perhaps the best person to define narrow i am but i also the way i try to look at Artificial Intelligence is through the ways that gen Product Designers and consumers engage with Artificial Intelligence. It was re here in doha if and when they get stronger is it a done deal that it will outperform us human beings well i dont lists and these customers of these idea of Artificial Intelligence for a while have been. Really over excited of the idea of an artificial being been able to take on a number of activities and the asks that we humans prefer not to do but i myself through my research i found evidence that makes me quite skeptical of the development of such technology in the short term and catalina coming to you when that happens how big an event how strong an event will that be so i very much alive it was just said the idea that were now looking at Artificial Intelligence as this blanket term that is supposed to really be everything and anything whereas actually and practice Artificial Intelligence is right now simply a set of tools that will perform these tasks that were mentioned or that were given as examples before hand so tools like facial recognition or tools and methods like facial recognition like the hate speech recognition and some of these tasks are fairing much better than others and for those that already do very well we really need to try to understand what particular Legal Framework is necessary and here its very important to keep in mind that since Artificial Intelligence is nothing new its been around already for decades we already have quite a lot of rules that can be very very fitting so if you go to your question. The kind of impact that these developments will have really depend on the accuracy of the methods and also the existing legal structures that we have in place caroline in berlin coming back to where are we with ai as of today i mean for you for example is facial recognition or worry because thats something that were reading im awful lot about now. Its a 100 percent its a worry advised for me in particular i look through it again through this lens of consumer and consumer understandability exam reproduction so how how Artificial Intelligence is folded into a product or Product Design visual recognition through i mean theres various studies out there the one i when i love to cite is one from mit by joyce balaam a gender shades which looks at amazon and some recognition i mean watson microsoft and i believe google as well and looking at how Computer Vision and facial Recognition Systems understand gender and then skin tone. Recognition but all of these systems but amazons recognition had a difficult time in recognizing darker skin tone so variety of different races and a harder time recognizing gender is performed best with white men and so when we think about like the the the pitfalls of facial recognition well imagine you know one the Emotional Trauma what it could feel like when this thing fails perhaps at a Border Crossing the more importantly what happens when there are failures what are the work around how that folded into a design what happens what is the extent of that harm and then going even further should we be using such error ridden technology on the general public and for the general public in Consumer Products so thinking about for for the end consumer. You know one of the harm of having something not be so accurate you know deployed at large to our users understanding how that system works and they understand and opt in or opt out is that explain to them or is it kind of off you skated went wrapped up into the product itself and then 3. How how like what are those you know again were these harms and is is it something that they can ever inside of a Service Design opt out so is it something that so a part of the flow that theyre in much like different Border Crossing if they were to. Try to not use the product or the Technology One can they is it possible and to what are the steps to avoiding it so i think its important sort of see where these products sit in the world at large and then how systems are designed around them so you know problems with facial recognition as users are we aware when its being used and are is there any potential for us to not engage with that if its being used in c. C. T. V. Cameras now right as an individual i cant have a certain c. C. T. V. Camera thats using facial recognition not work on me as an individual right so thats the way i think about is where the cues that user see what are you know how are these systems explained to them is it made legible what is legibility and how does design really to that and i think this is a bigger problem even outside of facial recognition but in a variety of products that use different kinds of Artificial Intelligence and its even raises a deeper question of do consumers understand the different kinds of Artificial Intelligence and then the Harms Associated with that and are those harms made idea that he boys rate here in doha what does it do or what will it do to the Global Workforce particularly if we can text julys that assertion in terms of say you know if youve got the workforce in china billions of people who work for very small salaries but if youve got to see a more sophisticated workforce manufacturing Something Else the impact might be completely different. Well but my question would be in return. Does this Technology Still allow to to have a more smarter more powerful workforce to some extend just to kind of bring it back to the area of my recent research on the field of journalism this we know is they believe that you know having Artificial Intelligence embedded in the newsroom will you know journalist from very. Time consuming tasks but and freedom up to do perhaps more Investigative Journalism for instance and in reality they stick knowledge as as they stand are unable to really produce a new story from scratch theyre still pretty much dependent on Human Authority sation and journalists writing templates thats the problem that is the problem to me is that Public Perception has been over excited with you know the promises of what Artificial Intelligence could could create of could produce but we are not still yet catarina go on to en mass straight just to go back to that idea that carolines innocent berlin was talking about if you have Automated Technology here im thinking about i have i have in my house automated an automated cleaner that comes out at 1 oclock in the morning if i program that to clean the house as fast as possible it might smash the coffee table it doesnt know what its done no harm done but if you have an automated drone aircraft dropping bombs the drone has anonymity and autonomy so the potential for catastrophe there is huge but the overarching principle of writing a program and instilling intelligence is much the same surely. Yes im very happy that youre bringing up all of these examples because theyre essential in understanding the differences between the harms that caroline was mentioning earlier the different harms that emerge out of different contexts so if youre speaking about Consumer Products so caroline was mentioning about facial recognition as maybe something that can be embedded i dont know an app like snap chat and actually it is already being embedded there but but that is very different indeed than having autonomy or automated highly or d partially Automated Systems that can be deployed in warfare so thats why from a legal perspective its very important to keep this context very specific because. The rules that normally apply also to the these fields theyre also very different so drones that attack Different Military forces that those are going to fall under or should fall in the future under humanitarian law rules whereas when were speaking about Consumer Products were also speaking about Consumer Protection and Product Liability and maybe to build on this idea but i think its very important for us to realize is that behind all of these industries and behind all of these markets there are Business Models that we have yet to fully comprehend im not sure how many of the audience members of todays show will be familiar with companies that do data and richmond because at the end of the day in order to deploy all of these technologies you just need data so that data harvesting approaches of kompany it can become more and more more and more negative so they are going to try to get information about consumers about to comp unease from many different places and by they i mean these data and Regiment Companies so this is something very vital that right now is not being completely. Lawrence and i think that we really need more evidence in this respect what are the data brokerage models what are the economic incentives and economic interests that these markets are right now. Creating caroline does that mean that we should limit the intelligence of Artificial Intelligence because of where it may end up im not quite sure how we could limit the intelligence of Artificial Intelligence systems given that Artificial Intelligence isnt really that intelligent i think what it comes back to is the intentionality and as my cocounsels mentioned data brokerage and i even go even further thank you about data sets and who maintains them and how open they are and the enemy in any of users within data sets its not necessarily about Artificial Intelligence it does actually come down to whats making up the ai system and in the context of where its use is so you know i think the previous answer was really fantastic answer separating Consumer Products that use automation versus weapons and war maybe at the very very very very basic level theyre using a similar kind of system or automation but in actuality theyre made in very different ways and theyre pulling from so many different kinds of things a big thing would be data set so the way a room but its design and the data set its trained on is probably very different extremely different than per se Predictive Policing Software right so i think its more important to think about the data sets and standards around what we consider fair or more balanced datasets like less biased datasets caroline thats all just to get used to the close partner for secondarily we have talked about data sets what youre talking about is the toolbox of human ask Decision Making the Artificial Intelligence utilizes when it has to take a decision on what it does next depending upon the process that its involved in at any particular moment. Somewhat yes so if i have a system using inside an Artificial Intelligence system you need data that data interacts with algorithms already our official intelligent code itself right and then its trained label model that cetera. And so its not just like it is these Human Systems it is also thinking about well what what is the problem im trying to solve and then what data set did i use around that so predictive policing for example uses historical police data right its using older arrest history but we can unpack how there are so many. Like in accuracies even Human Rights Violations within that i speak from a very particular we