A.I. + Designers (and pretty much everyone else)
An examination of the shifting role of Designers and other humans, as Artificial Intelligence firmly enters the picture
…A.I. is a concept to create intelligent machines that can simulate human thinking capability and behavior… [1]
Some will resist, some will get on board the hype train without a thought, some will cautiously venture and some will tilt their head skeptically because “it’s not there yet.” What started a few years ago as a blip on the radar of technology has started to turn some serious heads, and a lot of them. Designers are no exception to this; the implications of AI/ML in design are developing rapidly and we don’t really know what we’ll get even in the next month. This is exciting, terrifying, and awe-inspiring, all at once. We are in the eye of the storm, and it’s worth slowing down to enjoy the view.
So what’s all the fuss about?
OpenAI recently launched Dall-E 2, a system powered by GPT-3 that can create images based on text prompts. A new and improved version of Dall-E, this one is able to create imagery with staggering amounts of detail and artistic flair. It isn’t just compositing images from a database. It has character and creativity, often creating visuals with subjects doing things that no human, and hence no database, has probably ever seen [2]. Here’s a great video on why Dall-E 2 is incredible, and I recommend exploring the website too.
When I saw the website, I’ll admit that I was a bit rattled. It’s a frightening revelation that A.I. is able to create visuals with the artistry that most people I know simply don’t consistently have. Design is already firmly in the picture when it comes to A.I.; we already have a bit of an idea of how A.I. design works. This brings us to the question that is actually relatively young but feels age-old thanks to the rapid advances of A.I. and M.L.:
Are Designers and Artists really going to lose their jobs?
The short answer is no, they will not.
The caveat is that Designers will need to adapt and forge new relationships with technology [3].
Run for your lives, it’s the Big T!
Technology isn’t new to us- software has been a mainstream component of our discipline since the early 90s. If you saw a graphic designer who refused to work on Adobe Illustrator, you’d probably think they’re a red flag and wild card all rolled in one. The debate of relying heavily on software even exists today, at a time where Adobe Sensei is able to pull stunts with images and video that were unimaginable just 5 years ago. It is likely that designers will have a similar kind of learning curve when it comes to the boundaries and relationships between computer and designer. There will be growing pains and the new order will be forged in the heat of the uncertainties.
Jasmine Oh, in her article published on Microsoft Design, places this change in an actionable context:
… While A.I. will replace designers, it will replace the designers of today, not the designers of tomorrow. A.I. will become a design partner and tool that designers can use to meet ever-evolving workplace demands [4].
The creative field is poised for a few big changes in the next 20 years. I recently read an article on the Adobe Blog written by Linn Vizard on how the role of designers is changing vastly alongside the advent of new technologies [5]. She describes a new order of design which should strive to take a few steps back and look at the bigger picture, while allowing our A.I. companions and tools to carry out production of the details. This is an order that will slowly phase out space for what she describes as “Star Designers”, making way for Star Teams instead. Human centered design will be the name of the game, and this is already an ideology in which the concept of having a single glorified “genius designer” doesn’t quite fit in.
Where we earlier had multiple specialisations that created siloes within which designers practiced specialised crafts, we could soon have a more holistic design approach that seeks to design environments, behaviours, systems and experiences. The focus is moving from designing the image to designing the entire context and experience of which the image forms a small part. Designers and creative practitioners will likely need to go over and above to stay relevant. As Oh puts it, the Designers of Today will most definitely become obsolete, only to make way for the Designers of Tomorrow — those who are able to adapt and welcome this unified vision of design that A.I. could make a reality.
Hold your horses
Acknowledging that there have been staggering technological advances in A.I., Machine Learning and deep neural networks over the past 10 years, let’s also acknowledge the convolutions- for the purpose of this article, just one major convolution. A.I. Bias is here to remind us why there are a few reasons to hold our horses.
Bias in A.I. occurs when results cannot be generalized widely. We often think of bias resulting from preferences or exclusions in training data, but bias can also be introduced by how data is obtained, how algorithms are designed, and how A.I. outputs are interpreted [6].
This is an explanation of how biases occur in A.I., as told by Stanford cardiologist Dr. Sanjiv M. Narayan in this interview with Healthcare IT News. A.I. is a complex subject matter, so I’ll refrain from going into technical details (as I have throughout this article) and instead jump right into a few examples of A.I bias.
TW: References to Racism and Islamophobia in the next 2 paragraphs*
While researching for this story, I came across a Vox article on some problematic trends that have cropped up in A.I., mainly to do with Islamophobia. I recommend reading it but TL;DR: the language model GPT-3 (also used in Dall-E 2) has an overt tendency to regard Muslims as being inherently violent. The tweet embedded above is a video posted by one of the researchers, Abubakar Abid. You can see that he even tried to tell GPT-3 that the Muslims he’s talking about are peaceful folk, and GPT-3 would have none of it.
Google’s Photos app, that uses A.I. algorithms to label images and provide us with those often annoying “memories”, committed a pretty big blunder back in 2015 when it labelled an image of black people as “Gorillas”. Granted, it has been 7 years since this happened (wow, I’m older than I thought), but it’s a bias that is so blatant that one can’t help but wonder what on earth was used to train these algorithms. Apparently, Google has recently fixed this issue (not really) [7].
A final and less distressing example that I’d like to share with you is that of char-rnn. Situated in a corner of GitHub, this is a multi-layer recurrent neural network that writer-scientist-engineer Janelle Shane has been training on various forms of output. She got this neural network to churn out a series of pick-up lines — yes, the kind you use at bars and various other public spaces to acquire partners for romantic and/or sexual purposes [8]. The results were quite entertaining, here are a few:
I want to get my heart with you.
You are so beautiful that you know what I mean.
Your beauty have a fine to me.
You look like a thing and I love you.
The first thing that struck me about these was that they sound like a bunch of nearly nonsensical yet somehow profound sentences strung together by a child who doesn’t fully understand what they’re talking about. In this case, that subject is romantic and/or sexual attraction between two adult humans. The keyword here is child.
Baby shark, doo doo doo doo
We know that A.I and neural networks are created over time, developed, expanded and trained. One of the reasons why many of us are fascinated by this field of study is because there is an allure to this process, something that makes it seem almost human even though that’s exactly what it’s not. But I think that it reminds us of ourselves — we too, as children, grew up from being small humans with still developing brains. We grew slowly and over time, we developed, we learned, went to school, understood how the world worked, started asking important questions and working to find important answers. But A.I systems and neural networks generally have to start at the beginning and, as it turns out, there can be intense growing pains.
Janelle actually wrote a book about the intersection of artificial intelligence with modern-day technology and society, and it’s called You Look Like a Thing and I Love You: How Artificial Intelligence Works and Why It’s Making the World a Weirder Place (first published in 2019). I hope to read it soon and share my review of it, but don’t hold your breath because I read books at a laughably slow pace (hey, at least I’m trying).
There’s no overt indication that AI is going to destroy the human race and take over the world. In fact, here are 5 reasons why we needn’t and shouldn’t worry about that. AI Is being developed by many smart, big-brained folk around the world to help us, not to spell our end like Hollywood would have you believe. Take the Jaws Effect — Hollywood gave the world’s 1,000+ species of shark a horrible reputation and awful PR through the Jaws films, which painted sharks as a bunch of murderous monsters hell-bent on eating people. The movies taught people that “sharks intentionally bite humans, that human-shark encounters are always fatal, and that sharks should be killed to prevent future attacks.” What resulted was a culturally ingrained distrust, fear and hate of fascinating animal species that are an integral part of marine ecosystems. Following the release of the Jaws films, global populations of sharks and rays dipped drastically [9], an incidence that continues to threaten ocean ecosystems today. Oh, and by the way, “each year worldwide there are ~10 deaths attributable to shark attacks compared with ~150 deaths worldwide caused by falling coconuts.”
Not to draw a direct parallel with sharks and A.I, but Hollywood misleading people with theatrical mumbo jumbo is a recurring pattern that is highly avoidable if we could all just focus a bit more on reality. A.I is being researched and developed methodically and scientifically, showing us its wonders and capabilities along the way; the many ways in which it can make our lives easier, cleaner and more sustainable. Whether you’re a designer, a writer, a data analyst or even a restaurant owner, A.I. is coming to help you. It’s bringing us closer to a brave new world where there will be more opportunities and possibilities than ever before. It’s up to us to play our cards right and use this new technology for good, knowing what we know about humanities past mistakes with technology (and a few other things).
So buckle up and enjoy the ride with an open mind!
References
- “Artificial Intelligence and Machine Learning ; What Is the Difference?” Encora, 7 Mar. 2022, www.encora.com/insights/ifference-between-artificial-intelligence-and-machine-learning.
- ColdFusion. (2022, April 22). How This A.I. Draws Anything You Describe [DALL-E 2] [Video]. YouTube. https://www.youtube.com/watch?v=U1cF9QCu1rQ
- Philips, Miklos. (2018, July 10). The Present and Future of AI in Design (with Infographic). Toptal Design Blog. https://www.toptal.com/designers/product-design/infographic-ai-in-design
- Oh, Jasmine. “Yes, AI Will Replace Designers — Microsoft Design.” Medium, 9 Dec. 2021, medium.com/microsoft-design/yes-ai-will-replace-designers-9d90c6e34502.
- Vizard, Linn. “Technology and the Evolution of the Designerâs Role.” Adobe Blog, 24 Jan. 2017, blog.adobe.com/en/2017/01/24/technology-and-the-evolution-of-the-designers-role.
- Siwicki, Bill. “How AI Bias Happens — and How to Eliminate It.” Healthcare IT News, 30 Nov. 2021, www.healthcareitnews.com/news/how-ai-bias-happens-and-how-eliminate-it.
- Vincent, James. “Google ‘Fixed’ Its Racist Algorithm by Removing Gorillas from Its Image-Labeling Tech.” The Verge, 12 Jan. 2018, www.theverge.com/2018/1/12/16882408/google-racist-gorillas-photo-recognition-algorithm-ai.
- Shane, J. (2017, April 7). The neural network generated pickup lines that are actually kind of adorable. AI Weirdness. https://www.aiweirdness.com/159302925452/
- Salazar, Maria. “If You Think Sharks Are Scary, Blame Hollywood, New Study Suggests.” Mongabay Environmental News, 7 Sept. 2021, news.mongabay.com/2021/09/if-you-think-sharks-are-scary-blame-hollywood-new-study-suggests.
