By PAUL VON BÜNAU
Too often, ideation is performed without structure or scrutiny. We must learn how to do it well to avoid wasting precious resources and jeopardising people’s enthusiasm, energy and trust
Ideation sessions have become central to how many organisations think of new things. These gatherings are where many companies take their first foray into Artificial Intelligence and its seemingly endless possibilities. They’re where bright ideas are born, evolve and grow. Or, at least, that’s what’s supposed to happen.
It’s a sad fact that 95% of AI projects created in these workshops probably shouldn’t be launched. Many of the resulting pilot projects weigh heavy on resources – money, energy from misplaced enthusiasm and the precious time of senior leaders – and amount to little.
‘95% of AI projects created in these workshops probably shouldn’t be launched’
So why are so many undeserving ideas given these resources? It’s because ideation in AI poses a unique set of challenges. Here’s what they are and how your organisation can address them.
Lack of intuition
AI is a new technology and very few people know what it really is, let alone know intuitively how to use it and what it is capable of. With this lack of understanding, we can hardly expect people to come up with workable ideas.
The media has contributed to our lack of intuition too. Bad metaphors around ‘learning’ and ‘intelligence’ have created falsehoods and fantasies about what AI can do. This leads to worthless ideas being hatched, and relevant ones never being incubated.
How to address it
Every ideation session should start with an education session, in which capabilities are broken down into a small number of archetypes, such as ‘Filtering’, ‘Ranking’, ‘Searching’, ‘Information extraction’ and so on. Hold one of these interactive education sessions with practical exercises before you get to ideation. If possible, have experts in the room for when you need extra detail or questions answering. But don’t let them take over the idea generation. Too much detail and not enough free spirit may weigh heavy on the session.
‘AI is a new technology and very few people have developed an intuition on how to use it effectively.’
Misplaced focus on data
Starting from a data standpoint has undoubtedly sent many ideation sessions in the wrong direction. I understand why this happens. Data has been (wrongly) heralded as the new oil, and while it is indeed a key input to most AI systems, it is not the driving factor.
If you’re overwhelmed by endless possibilities during ideation, anything that limits the scope can feel welcome. But be wary. The data you currently have could be a poor guide. It could limit the potential of ideation, and rule out the most strategically relevant, long-term applications of AI – which often rely on adapting processes to capture data you don’t yet have.
How to address it
Frame ideation sessions well and ensure they’re structured to begin with business objectives and purpose. Focusing on purpose can also help to focus minds on more than just the data you have available to you.
Only at the very end should you arrive at landscaping for data. In this step, make room for data that doesn’t yet exist. Is it possible to capture it? If not, which third party might have the data? Explicitly encourage people to think beyond the status quo.
We humans enjoy navel-gazing
Thinking outside the human mind is tough. We’re often misled by anthropomorphising metatapors. In many cases, we’re still naively trying to put AI where people were before. And in doing so, we’re ignoring AI’s and humanity’s specific weaknesses and strengths. Instead, we should be more thoughtful around ‘where to play’. Take the example of sculpting. Human ingenuity is still unmatched when it comes to creating sculptures. Computers on the other hand, do not tire and a 3D printer could create a thousand sculptures in the time a human honed a dozen. Our strengths lie in different places. We must remind ourselves of this.
How to address it
Hosting an education session before your ideation session is the way to go here. Confront this issue head on and use concrete examples to demonstrate the case at hand. Ask participants to rate tasks like ‘Scanning two million images for the occurrence of screwdrivers’ vs. ‘Checking a single legal contract for logical errors’. Boil it down to simple conceptual rules-of-thumb like ‘when it’s fine to be n% correct on average’
Asking the wrong questions
To later separate the wheat from the chaff, you’ll want to look into many different aspects of your ideas – with a keen eye on ROI, feasibility and risk. Yet where to scrutinise and which question to ask might not be that obvious.
How to address it
Use a canvas to help you. Here’s ours. At the same time, don’t think the canvas will do it all for you. You still need a couple of expert guides to help participants with any missing detail. For the follow-up not to get bogged down by lots of missing information, make sure you ask all the right questions whilst everyone is still in the room.
Bringing in IT too early
AI looks deceptively like just another form of IT. Yes, it runs on IT. And the ultimate success of every AI project critically depends on an appropriate IT infrastructure to support it. But in the beginning, it’s all about pinpointing where AI can add business value, and not about discussing implementation details prematurely.
How to address it
At first, frame the workshop as not being about technology. Similarly to how we covered our data challenge, you should focus on business objectives, purpose and value and not the limitations of IT or the infrastructure you have already. Address the big picture before filling in any feasibility gaps. And make it clear when and how IT will be involved. For example, in certain subsequent steps when assessing feasibility.
With some change, ideation in AI becomes a powerful tool
Despite its challenges, ideation in AI undoubtedly still has its place. Many of tomorrow’s ideas will be launched from ideating together – whether in virtual sessions enforced by the pandemic or face-to-face gatherings.
I believe a little more scrutiny and structure in ideation could make these sessions far more likely to yield success.
Download our ideation canvas here.
Once you’ve hosted your ideation session and your AI project ideas are beginning to take shape, the next stage is to do a thorough opportunity assessment. Get in touch if you need our help.
The Economist’s Technology Editor, Tim Cross, covers how people have perhaps been a little naive in misunderstanding AI’s limitations. This touches on point three in our article above. The author questions whether the hype many subscribed to has left some believing AI has failed to deliver.
Looking beyond the lockdowns, face masks and hand sanitiser, the pandemic made many people look towards technology as a savvy defence mechanism against COVID-19. Here, the Financial Times writes how AI has disappointed all of us. Could better ideation have helped make progress? You decide.
Here’s a detailed article by Google focusing on how to design for AI. As we touch on in point three above, it covers how to bridge the gap between human expectations and AI’s abilities. Their Chapter about ‘Mental Models’ gives advice on how to re-shape people’s expectations on AI to form realistic impressions.
The Apollo 13 Mission Control team faced a huge number of seemingly insurmountable obstacles after an oxygen tank exploded on board the 1970 mission to the moon. How did they solve it? Ideation. Here, Design Better’s article about Brainstorming illustrates why ideating openly with others is crucial for fast and promising problem solving.
How can our biases negatively influence brainstorms? This article suggests ideation sessions should have rules, but that the rules should change dramatically once AI gets involved. It covers humanity’s complicated relationship with AI and touches on many of the themes in our piece.