Is artificial intelligence (AI) going to take your design job? Well, certain tabloid newspapers certainly want you to think that 52% of all jobs could be taken by ‘robots’ (which, please note, aren’t the same as AI) within the next ten years. It’s true that AI already exists and is already functioning in our world. It is helping farmers kill weeds without harming or polluting crops. Image recognition AI is helping look for signs of diabetes in people’s retinas. AI is beneficial to detail-oriented jobs which require accessing and filtering a vast amount of data.
But that isn’t going to put half the population out of work.
Even so, Elon Musk – one of the world’s brightest and richest men (and founder of his own AI company) – is quite animated about AI, saying that we are “summoning the demon”? Is he over-reacting? After-all, Facebook uses AI to target advertising at its users. AI powers digital assistants like Siri and Alexa. Indeed, Google’s search engine has been dependent on AI since day one. They’re all good things… Right?!
Have you trained the autosuggest function on your phone to predict the types of words and phrases you’re likely to use when texting? That’s AI.
When you’re working on your computer in a platform like Google Docs, does it accurately suggest corrections to your typos? That’s AI.
Does your streaming service suggest playlists that nail your taste in music? You guessed it: AI!
These are but a few decent things that AI offers, so why are some people scared?
A lot of this thinking comes from the way AI (or ‘supercomputers’ as we used to call them) have been portrayed in the media. Watch a science fiction film and you could be forgiven for being afraid!
On the trAIl of AI
Anyone who watches science fiction movies has been aware of AI for some time. Robots in TV shows have been ‘intelligent’ for decades, from the unimaginatively named ‘Robot’ in Lost In Space through to the diminutive Twiki in Buck Rogers in the 25th Century and the Pinocchio-like Data in Star Trek: The Next Generation. The Star Wars movies, famously, gave us the AI Laurel and Hardy double-act of R2D2 and C3PO.
These AI robots were generally seen as mostly benign, possibly because they look (vaguely) human. The way non-humanoid AI has been seen in movies, however, is very different.
Let’s take Hal 9000, for example. The film 2001 never really bothers to explain why HAL goes mad, but the novel tells us that he was programmed to always tell the truth and, at the same time, to not tell the astronauts about the existence of aliens. HAL decided that the simplest way to successfully do both these things was to kill the astronauts.
When these computers became really scary, however, was when science fiction filmmakers imagined them taking over from we poor, illogical humans. In WarGames an AI computer was given control of our nuclear defence arsenals and promptly turned them against us, because it thought it was playing a game. It did, eventually, realise that some games can’t be won so shouldn’t be played, and decided it would prefer a nice game of chess.
SkyNet, the AI in the film The Terminator did not come to the same conclusion. It feared being switched off by humans, so protected itself by creating Terminators.
Do you want Terminators? Because that’s how you get Terminators!
So, it’s fair to say that AI has had a bad press… But let’s take a breath, calm down, and look at what AI really is.
Giving AI a fAIr hearing
Not so long ago (within the lifetime of many designers still working), Graphic Design was a job done with pencils and rulers, scissors and glue. Then the Apple Mac led the way to taking all (well, nearly all) Graphic Design onto the screen. There were just so many more things that could be done at the click of a mouse.
Where AI can help with on-screen design is to quickly put together a range of suggestions, whether that be in product design or web design. These suggestions can be assembled from the available elements, employing the skills and rules the AI has been taught. The latest artistic composition tools allow designers to create sketches, and the AI will contribute existing completed assets that resemble or work with the sketch, live, while the composition is taking form. This can help speed up the design process by helping creatives avoid reinventing the wheel. But a client still needs to okay the design. They still need to like it and, as any creative who has ever worked with a client will know only too well, there’s no accounting for (lack of) taste.
Effective websites are designed for two audiences – search engine bots and human clients. A webmaster has to use all the well-established techniques of web-design, coding and SEO to attract the attention of the search engine spiders, but that is not an end in itself. The search engines bring humans to your site. That is the endgame! The content needs to be relevant to those humans, and that’s something that only another human can truly decide.
Similarly, when you are posting your content on social media, images need to conform to established and measurable rules, such as image size, file format and colour models; but the creativity behind the image, the strategic thinking that will make people stop scrolling and engage, that’s all human.
Hopefully you will notice that, in all of these applications, the AI is doing the research, it’s assembling data or assets that it has found, it is not creating. The truth is, a lot of the work of building a website – for example – is not creative, it’s about following rules and paying attention to tiny details. These are things that computers are very good at because, unlike humans, they don’t get distracted or bored.
They also don’t create.
WAIting for AI
AIs do not (yet) have ideas, they just assemble information that they have been given by human programmers. And they don’t yet do it perfectly, so they can’t be set unsupervised tasks.
In that sense, AI is still growing up!
There are limits to machine learning, just as there are limits to human learning. It takes time. There is a Cognitive Psychology experiment in which a scientist programmed an AI to recognise a panda bear, by checking for the following features:
- Black and white colouration.
- Prominent ears.
- Sitting upright.
- Has fur.
- Is an animal.
In this way, the AI algorithm successfully spotted a panda pretty much every time it was shown one. Then, the scientist showed the AI a second picture. The algorithm checked its list and confirmed that this was, indeed, a panda. However you, being a human being, can glance at the picture below and tell instantly that this is an incorrect deduction.
This happened because the computer was given limited information with which to work. It needed to be given more data about the pandas, it needed to be told how to spot the difference between a real panda and a pretend one; it also needed to be told how to cross-reference with its check-list for recognising a baby. This is all information an AI needs to be given; unlike humans, it can’t go out there and learn by itself.
Gary Marcus, who was the head of AI at Uber and is now a professor at New York University, wrote an academic paper called Deep Learning: A Critical Appraisal, in which he details the ten areas he believes AI has limitations. One of the key limitations is the vast amount of precise data it needs, and how poor even the most sophisticated systems are at inferring from context. They just can’t improvise.
That fact alone, that lack of imagination, should give people working in the creative industries cause to relax.
The future of the media
It has been suggested that AI would be less scary, and better understood if we redubbed it Augmented Intelligence or Assistant Intelligence. Yes, AI smart tools might be able to quickly assemble elements into a variety of designs, but it can’t tell which of those designs is actually aesthetically pleasing. Designers still need the mark one eyeball for that!
So, AI can enhance what designers and creatives of various hues can do, it can speed the process up. But it will be hard pressed to replace you.
As Silka Miesnieks, head of the design lab at Adobe put it: “Our superpower as humans is our creativity, and we are all creative … AI can aid human creativity but not replace it.”
Rod Farmer, digital expert associate partner at McKinsey & Company added: “In most cases, AI only affects a percentage of a job a person does, but rarely the whole job. And if you free up workers to do more of the creative and other work they are good at, while leaving other tasks for AI, you can grow your organisation.”
So, far from taking away 52% of people’s jobs, AI will free people up from the part of their jobs which is uninspiring and repetitive, allowing us all to concentrate on the fun bit – the actual design work!
Now, how about a nice game of chess?