Fri, 19 Aug 2016 16:51:55 GMT
Hollywood has been obsessed with Artificial Intelligence for as long as computer based technology has existed. At times, we're presented with the idea that robots are cute, almost cuddly, things that we should have nothing to fear from – think C3P0, WallE and the tragic but lovable Marvin, the Paranoid Android.
However, most of the time, AI is treated with suspicion and fear – think the Terminator, GLaDOS and that softly spoken psycho, Hal 9000. Issac Asimov’s Three Laws of Robotics, the ones where robots can’t go around maiming and murdering us humans, are more often than not bypassed completely in favour of the human race getting royally done over.
But, in reality, do we have anything to fear from a robot dominated future? How long will it be until the AI androids surpass human creativity? And what the hell is deep learning?
For Mike Dory, Managing Director of Technology at KBS and creative-technology boutique Spies & Assassins, throwing around AI terminology can be an absolute minefield. Dory explains, “When people start throwing around those big words in discussions about AI, that's usually when you know you're talking to someone who doesn't really know what they're talking about.”
So, with Mike's warning ringing in our ears, let’s tick off a few key terms and phrases we all need to learn before our machine overlords take over, because one things for sure, when R2D2 and his other pedal bin pals turn on us, I want to know everything there is to know about their stupid little robot brains.
This is basically when we’re all fucked. Singularity is the moment when technology will have matched the creative capacity of humans and lead to an “intelligence explosion resulting in a powerful superintelligence that would, qualitatively, far surpass all human intelligence”. Gulp.
This is the view of Ray Kurzweil, author of the New York Times bestseller The Singularity is Near, and artificial intelligence expert. Kurzweil expects singularity to occur around 2045, so that's under 30 years away. A generation.
But does this mean that we should just give up? Will machines just steal our jobs and leave human creativity behind? For Dave Cox, former Chief Innovation Officer of M&C Saatchi, Kurzweil's prediction is a critical one. “Once we reach that stage, all bets are off,” admits Cox. “But I think it’s interesting that people are scared of AI, because it will eventually be so much smarter than us. Of course, we should monitor its progress, but right now its ability to think at human creative levels is way off.”
Cox and the team at M&C Saatchi designed the world's first AI billboard. A smarttech advertising billboard that was given free rein to create adverts for a made up coffee brand, the results of which M&C Saatchi later studied. However, despite the incredibly clever tech behind the billboard, most the work was still done by humans. Cox continues, “Much of what the AI billboard created would not be considered ‘good’ advertising, but it certainly got a lot of attention so you could say it was therefore successful.
“It would take a brave client who is able to accept the funny consequences and results to use it. So, in my view, AI is still not at a level where Creative Directors should be worried about their jobs.”
M&C Saatchi certainly learnt that, despite Kurzeil's warnings, billboards and ads will still, for the moment at least, be best utilised in the hands of humans. “The key will be not letting the AI actually drive forward the creative,” explains Cox. “Rather, just let the algorithms optimise different things – like changing the ad when the weather’s different, the logo size or other minor details.”
Ahh yes, algorithms. What are they again?
Well, according to Mike Dory, “Algorithms are the fundamentals; It’s how all AI systems understand the rules.” Basically, they are the bedrock of anything artificially intelligent. Recently, the growth of algorithmic complexity has led to some pretty cool developments. In Japan, the spiritual home of all things tech, algorithmic genius was at the heart of the creation of the first ever AI Creative Director.
McCann Japan Millenials with AI CD ß
Given the catchy name, AI CD ß, the android Don Draper was created by McCann Japan’s ‘Millennials taskforce’ as part of their ‘Creative Genome Project’. Shun Matsuzaka, Creative Planner of McCann Japan, and founder and leader of McCann Millennials, explains, “It is generally believed that artificial intelligence cannot 'create' or be effective in the creative field; however, lots of entertainment is already being produced by algorithms.
“AICD β is the world’s first artificial intelligence that can give creative direction to make commercials. To build the AI, we looked at some 1,000 award-winning TV commercial works from the past 10 years from the most authoritative award show in Japan, the ACC, and tagged them with original tags. We then broke down and analysed them to find out what makes a good TV commercial installed as its database. Once the creative brief is input, based on the industry, objective, communication content, the optimum creative direction is derived by the algorithm.”
Basically, much of the 'heavy lifting' still has to be done by humans. But the beauty of McCann Japan's algorithm is that it is always getting smarter, as Matsuzaka continues, “In the same way as AI’s active in Hollywood movies and music scenes, AICD β is based on proprietary tags developed and tagged manually by advertising creators. Data of over 1000 TV commercials have already been input. By increasing, the data volume, Deep Learning will be realized.
“The AI will not only refer to its database to give creative directions that will produce the most effective commercial for that particular product category, but it will also evaluate and gain learnings from the result after the commercial is aired so that its precision will be improved on future projects.”
Impressive and scary in equal measure. But how do machines really learn?
Machine Learning/Deep Learning
“Deep learning,” claims Mike Dory, “is the AI concept that is applied most freely. Most people like to think that everything can be done by AI, whereas in reality it remains just a tool, rather than a panacea.”
For Sheldon Monteiro, Global CTO of Sapientnitro, AI is a tool that companies can use to get rid of the tedious jobs humans hate. “We're not just talking about chatbots and things like that,” says Monteiro, “AI and machine learning technology today can be used to really help businesses become a lot more streamlined and efficient. If there's anything that involves routine work by humans, you can teach an AI system to do it.”
One such example of a business buying into the power of machine learning and killing off tedious tasks, is Persado. By inputting thousands upon thousands of words, phrases and images into their machine, collected from years of data mining across campaigns for companies like Vodafone and other giant brands, Persado has created a, “Cognitive content platform that combines natural language processing and machine learning technologies to generate the precise words, phrases and images that can inspire any given audience to act, every time.”
In technophobe speak, Persado's software can tell you what the best possible message for any campaign will be. The firm is bold enough to claim that their “cognitive content outperforms man-made messages 100% of the time.”
Lawrence Whittle, Chief Revenue Officer of Persado, adds, “Once you understand what triggers individuals emotionally, you can sell them anything. What Persado is able to do is create the best possible message by analysing the success rate of previous messages across a huge spectrum of data. We take subjectivity and add in objectivity through humanassisted machine learning”
With Persado, the potential for growth is virtually limitless, as the machine is constantly analysing new messages and narrowing down towards the best possible result. Whittle continues, “We have to admit that at times machines are better at certain tasks than humans. However, although Persado can find a better result than most humans, ask a machine subjective questions and they're often stumped.”
So despite everything machine learning can achieve, ask a robot if a Wagon Wheel tastes better than a Freddo, and they'll reveal their robotic innards. Winning the Imitation Game is still some way off for AI.
Saved State (Conversational User Interface)
Although we don't always realise it, many of us carry around AI in our pocket. Siri, the iPhone’s intelligent personal assistant, has come along way from it's initial release four years ago. The hilarity of torturing Siri may now have worn off (I can't remember the last time I asked Siri to call me 'Tits McGee'), but the AI has improved massively and more complexity has been added.
Saved state is an advancement that has added a new clever dimension to AI like Siri. “Certain bots are now able to tell you more about yourself, rather than just telling you what the weather’s like or what day it is,” says Mike Dory.
Siri and other bots can now understand the connection between an initial question and a connected followup. As Dory explains, “Computers can contextualise more readily and understand that questions, and therefore answers, can be linked. This is also known as a conversational user interface.” Although this may seem like a small progression, Dory believes it’s “real progress.”
However, despite Siri evolving into a much smarter personal assistant, it can still be confused pretty easily. But Dory says this is not really an issue as, “Bots and AI are now embracing their nonhuman side. Machines failing to be human is creepy – unless they embrace their failure, and then it's pretty funny.”
So will Siri pass the Turing Test any time soon? Well, no. As Dory concludes, “Most bots are still far from human levels of intelligence. And you can still confuse them pretty easily.” Saved state is progress, but Siri is still no smarter than the average toddler.
So are we doomed or not?
Well, according to the current zeitgeist in the creative and advertising industries, we're fine for now. “Most breakthroughs in tech have a great social benefit,” claims Persado's Whittle, “but there is a dark side we should monitor. But if you look at medical discovery, for example, AI can make very quick decisions that humans simply cannot, and that can be a huge benefit in things like medicine. For me, the reality is that AI benefits humans 99 times out of 100. Of course, it's topical and exciting to talk about Skynet and the Terminator, but in truth AI will help humans more than it will harm.”
Donald Chesnut, Global CCO of SapientNitro, believes the whole concept of AI is skewed. “The term AI is a weird one, it's not really that useful and it's got so much baggage. It's just simple nomenclature. We've got an opportunity now where machines are getting smarter. It's not about replacing people, it's about repositioning people to where the value is.”
Mike Dory tends to agree, and feels the promises made by Tomorrow's World have come up disappointingly short. “The past's vision of the future is usually way off – like seriously, where is my jetpack?! What we have now may not be 'real' AI, but I guess a lot of the tech around today is still rather futuristic. Things like the Roomba could have come straight from a SciFi flick from my younger days. Overall, however, AI will continue to improve, but it's taking time. There will be incremental changes and advancements, but for now we have nothing to fear.”
Much like Brexit, dabbing or catching a wild Snorlax in a ditch, the rise of AI, and therefore the fall of mankind, has been a hot topic this year. Elon Musk, who sounds more like an aftershave contestants on 'The Apprentice' would dream up than any kind of visionary AI pioneer, has claimed that artificial intelligence is our “biggest existential threat”; Stephen Hawking has warned that AI could end humanity; and with films like Her, Chappie and the fantastic Ex Machina achieving huge success in recent years, Hollywood's infatuation with artificial intelligence shows no signs of fading away. But in reality, it seems that the end of human creativity and our eventual enslavement by some kind of army of Apple androids is some way off. Which is exactly what they want us to think.view more - Trends and InsightLBB Editorial, Fri, 19 Aug 2016 16:51:55 GMT