Will the universality of AI knowledge make every space look the same?
Use of AI in the design of environments is currently hit and miss, but in the near future access to verified knowledge will improve. At that point, designers will face difficult ethical choices
The word seems to be getting around about what AI really is. It’s the tech version of cheating off the test paper of the person sitting next to you – who may have studied for the test, or who may not have done so. Also, that neighbour’s handwriting may be difficult to read, so it’s hard sometimes to even know what the words they’ve written down mean.
Most people using AI to research design-related topics have learned that they need to check every line of info they receive to make sure studies noted exist; and if they do exist, if what the AI is reporting aligns with what the studies found. Much of the best research remains behind firewalls, so what AI has to work with is – at best – somewhat limited.
Things will change
Making an error with the restricted bits that can be scraped off the open-to-all sections of the web can seem particularly heinous, especially if a client company will spend a fortune on design as a result of what’s presented. But someday things will change. AI will become less like the person sitting next to you during a test and more like a revered sage, and then we will have some difficult choices to make.
In the probably not too distant future, AI will somehow have snagged access to all the findings from the studies in peer-reviewed journals behind firewalls that are acknowledged by experts in the design research field to be of value.
AI will have learned, somehow, not to incorporate all the material it can find on a topic; quality control will come to AI. Musings that your Aunt Helen has posted online about how colours influence her mood will not be used by AI in the same sentences with findings from studies from the major colour research centres in the world, who have rigorously studied colour-mood links for decades, building up a powerful body of knowledge.
An ethical issue
It will be challenging for AI developers to make sure that AI uses only meaningful data in its reports and analyses – and that difficulty means a future timeline is murky at best. The only thing we can be certain of is that someday soon AI will grow up and be a more valuable resource. Which is when ethics will come into play.
Once everyone in the world with access to a phone and a charger can know everything we’ve ever learned about how design influences human wellbeing, mood and performance, what code of ethics will ensure that design solutions reflect all this information?
This ethical issue will not just arise in the design world, but in all other fields with specialised knowledge bases whose application can influence human welfare in some way. And this information will come to people who understand how to interpret, apply and integrate study findings and those who do not; this makes it easy for charlatans to claim to have reviewed findings in the greatest of detail and developed shortcuts to euphoria for people without the time to do so.
Testing the limits
So, how will professional codes of ethics, both formal and informal, evolve in the face of all of the information that will become available? If somewhere in the AI universe, for instance, accurate information is available about pain tolerance when certain sorts of art are in view (and this research does exist), and an interior designer does not specify this art in a hospital treatment room, will they lose their job? Will they face prosecution?
When knowledge is universally available to all, what are the consequences of not doing so for the professionals who can actually use it? Would the needed ethics code result in all designed spaces seeming the same? Will the AI end up dictating a narrower range of acceptable design decisions?
Most of the research in the neuroscience-informed design field provides general guidelines but does not dictate exact choices. In the example with pain tolerance and artworks in hospital spaces, the research points to certain sorts of scenes as being best, but not images of a particular meadow in Cornwall, for example—but would an AI ethics code be flexible enough to allow such variation?
It is questions like these that test the reasonable limits to what can be asked of well-intentioned professionals when ‘all’ knowledge is, at least theoretically, available to them through AI.
