As artificial intelligence (AI) develops and robots become more sophisticated, this is naturally causing trepidation as well as excitement.
While Black Mirror episodes are depicting AI robots as husband-replacements or entire species ready to wipe out the human race, the more rational fear seems to be centred around how robots could one day take our jobs.
Could robots be programmed to have an imagination?
Automation is happening already – Amazon has recruited a fleet of robotic assistants to shift stock backwards and forwards in its warehouse in Manchester, while self-service kiosks grace the likes of Sainsbury’s, Tesco’s and McDonald’s across the UK, reducing the need for cashiers. The BBC even launched a tool in 2015, which allowed people to test how likely it was they would lose their job to a robot in the future.
Unsurprisingly, creative jobs such as designer, artist and author, come up as pretty unlikely. The argument has been that, while robots can be programmed to complete repetitive, logical and intellectually difficult tasks – could they ever be programmed to feel? After all, creativity requires emotion and imagination as well as logic.
Microsoft’s bot draws what you tell it to – and what you don’t
New software from Microsoft challenges this notion. The company has developed an AI robot that can draw an entirely new image based on whatever a human tells it to draw. Rather than operating like a Google Search and finding an existing match, it will actually generate an entirely new image based on specific words.
Perhaps most scarily, it can also fill in the gaps. If a person asks Microsoft’s Drawing Bot to generate a photo of a yellow and blue bird with a short beak sat on a branch, but does not mention the background, the Bot could create something itself – implying that this AI bot has some kind of “artificial imagination”, according to Microsoft’s researchers.
The Drawing Bot, which seems far-fetched, was actually developed using two forms of existing tech: software which automatically writes photo captions for images, and another piece of software that has been used to help blind people by answering humans’ questions about images, such as their colour and object placement.
As clever as it is, the bot can get confused
Microsoft’s research team does admit that it is possible to confuse the Drawing Bot – simple text descriptions such as “blue bird” will be much more effective than a very detailed description, which could result in a blurry, generated image.
Equally, absurd descriptions can perplex it because it has nothing from its “memory” to compare to. According to the researchers, asking for a “red double-decker bus floating on a lake” in testing resulted in a mish-mash image of a monstrous, bus-boat amalgamation, suggesting an “internal struggle” in the Drawing Bot’s understanding. Perhaps its imagination isn’t quite as sophisticated as a human’s yet, then.
While it’s quite hilarious to imagine the object combinations that would most confuse and trouble this robotic artist – “draw a man with a pencil for a penis” or “draw 8,000 ant-sized elephants swarming towards an elephant-sized ant”, for instance – the scope of this tool is quite frightening.
Help or hindrance – what do designers think?
The researchers intend it to be used as a “sketch assistant” for painters and interior designers, or even replace the manual work put it in by a graphic designer or animator altogether. While the tech is currently imperfect, as the Drawing Bot builds up its knowledge of images, could it eventually become more imaginative than a human? And could this put illustrators and designers out of a job?
We asked designers for their opinions. The Drawing Bot is currently under development in Microsoft’s research labs.
“Although it’s very sci-fi to read about AI bots rendering a perfect image based on a caption description, I have my doubts that something like this can ever truly replace the creativity, combined with experience, that human touch can provide.
Ultimately an AI can only draw from existing content to create visuals – a human mind has the ability to imagine and create things are totally new and unique.
If you asked 10 people to draw (either digitally or physically) a yellow bird with black wings, the beauty of the results would be that no two would be the same. Perhaps this will become a new tool for creative minds to apply and use within their fields, as opposed to a rival.”
“While, on the face of it, the Microsoft AI drawing tool looks incredible, I have started to view headline-grabbing AI innovations with skepticism.
In my experience, these tools often appear to complete tasks better than humans – until you realise that the parameters in which they work are actually super limited.
I’m not an AI sceptic and there will come a time when machines can perform any task better than humans. But the common perception is we’re talking within the next 100 years – not the next five, or even 10.
I’m reading Max Tegmark’s Life 3.0 at the moment, which is great for anyone who wants to understand AI better. It portrays a future where machines start a media business, creating animated content and short films that are so successful that people prefer them to original, human content.
It made me rethink the outdated view that computers will never be able to think creatively. Creative thinking, imagination, and storytelling have always been considered uniquely human traits – but maybe we’re underestimating how computers can progress. Inversely, perhaps the way we learn as humans is actually quite machine-like.
Creative AI is coming – but I think it’s for designers’ children’s children to worry about, rather than us right now.”
“This does look like fun, but I can’t see it being a threat to illustrators in the grand scheme of things. Surely this bot can only create in the style it has been programmed to – unlike an artist, it can’t evolve its style or conceive a new illustration style.
It can put a bird on a branch, but can it illustrate the difference between a joyous springtime bird on a branch and a threatened, weather-beaten bird on a branch? Can it convey a mood or a story? It might be good for soulless, stock-imagery creation – and if so, good luck to it!”
“The Microsoft Drawing Bot is interesting but it still requires a human to tell it to draw – so we’re not about to enter a dark age for the design industry just yet.
I’m more worried about the conversation around Artificial Super Intelligence, which will be able to rapidly improve and reprogramme itself for 24 hours a day, seven days a week and on millions of computers simultaneously.
Once this happens, we won’t be able to keep up with the speed and evolution of thinking.
On a positive note, as AI becomes more capable, it might force creatives to be more original, in an age where we’re constantly bombarded with design inspiration from blogs and social media.
But really, speculating on design industry impact is probably irrelevant given that Artificial Super Intelligence threatens civilisation as we know it!”
What do you think of Microsoft’s Drawing Bot? Let us know in the comments below.