Robots Can’t Tell Jokes! Six things I Learnt about AI

by
Geoff Bilbrough

Agreeing to moderate a panel on a topic you know only a little about is a great way to learn and quickly. What better way to accelerate up a learning curve than being on stage quizzing the experts! Here are six things that I discovered moderating a panel on “The Future of Communications in an AI World” at the 2017 In2 Innovation Summit in Hong Kong.

  1. It’s all about the learning

    Computers do all sorts of stuff for us automatically already, from those annoying ads in our Facebook feeds to autocorrecting what we type on our phones, but what distinguishes this from artificial intelligence (AI) is the learning. With AI, the computer learns and gets better at what it does without the intervention of a person. Doing the same thing repeatedly is what computers do today; learning and improving is what creates AI. It could be said that ‘machine learning’ (another widely used term) is a better description of the all-pervading trend that is upon us.

  1. AI is here already

    We are already interacting with increasingly intelligent machines. It’s probably a ‘bot’ of some sort and most likely a ‘chatbot’. Next time you’re on your internet banking site and you get an invitation to chat, it’s likely to be a bot programmed to answer the simplest, most obvious and most frequently asked questions. The finance world has ‘robo-advisors’ – robots that advise you how best to invest your money and can trade on your behalf too. In the US they have bots that negotiate deals or file Class Action suits for you without a lawyer.

  1. People make mistakes, AI makes less

    The sci-fi staple of robots going rogue permeates all discussions about AI. But as it was tactfully pointed out by one of my panellists, humans working in call centres all over the world make mistakes every day that can and do cost money. Chatbots do too. However all the evidence suggests that bots working in the same environment as a human will make less mistakes. And given they are learning all the time and not forgetting what they learn (unlike us), the error rate will continue to decline.

  1. Bad data equals bad AI

    Meghan Trainor may think it’s all about the bass, but really, it’s all about the data. Data is the fuel that powers AI: no data, no AI; bad data, bad AI. Because of this it’s vital that data security is improved. Hackers stealing and reselling credit card numbers is one thing, but in the world of AI, people hacking into a database and changing the data behind an application is a very scary prospect. Tesla just changed the range of their cars remotely to help people evacuating from the recent hurricanes that wreaked havoc in Florida. You don’t have to think long or hard to come up with a similar but much more sinister scenario at the hands of hacker.

  1. Kids: get jobs in data science

    Jobs that are repetitive and process-oriented are at risk. The estimates of potential job losses in various industries are huge and, of course, speculative. That aside, there is no doubt that AI can do some of the tasks we do every day, reliably and to a high standard. Robots can make cars on production lines, pick and pack items in warehouses and increasingly they can chat with us on websites and help us make decisions. Work is changing and will change exponentially in the future. And as a panellist so kindly commented, communications is not immune. But while ad campaigns may be being developed by AI in Japan, the immediate focus has to be the repetitive tasks we face every day. This is where we can and will apply the machine learning power of computers in the future.

  1. Robots can make music

    It is true that robots can’t tell jokes (yet) but in my research for the panel I discovered they can make music. Magenta is a Google Brain project to ask and answer the question, ‘can we use machine learning to create compelling art and music?’ Have a listen to what they have created. As they say on the site, “this isn’t a performance of an existing piece; the model is choosing the notes to play, “composing” a performance directly.”

On the panel we talked and answered questions for nearly an hour, at the end of the session we all knew that AI is remaking our world.