The industry is falling over itself to make products ‘intelligent’ but what does that really mean to a user? As a Product Manager in this space, I’ve been thinking about intelligence and what it really means to users. There are many definitions of intelligence but they tend to be technical definitions of AI rather than intelligence seen through the eyes of a user. As we progress through the stages, what users once considered as intelligent, can become ordinary over time. For example, improving search so that it was accurate and fast could be regarded as highly intelligent at the time but now it’s the minimum we expect. As we move through technical advances in AI, the bar becomes ever higher. We now expect intelligence to transform our lives in a more drastic way than ever before.
Intelligence over time spans Knowledge, Understanding and Assistance as we have available today, to more human like attributes such as Complexity, Creativity, Emotion and Self Awareness.
Right now intelligence is optimized for Assistance, the combination of Knowledge (access to the world’s information) and Understanding (recognizing queries through natural language, and other machine learning) is a perfect application for Bots and Digital Assistants. Soon we will move into Complexity with the onset of self-driving cars and specialized assistants. If we go a little deeper on the definitions of these intelligence stages, we can understand where we could design products to appear intelligent to users.
Knowledge
Broad and in depth information i.e. Knowing about a lot of things is the traditional form of intelligence we are familiar with. The web has given us access to this huge range of information that makes us appear more intelligent.
Understanding
Today we have many ways of understanding what is being communicated e.g. Voice recognition, Video (body language and emotion), Audio (translation), and Photos (computer vision).
The intelligence is the machine learning that powers the recognition and understanding. This area of intelligence is improving fastest right now with error rates down to impressive levels e.g. Microsoft recently demoed the lowest error rate for speech recognition at a rate of 6.3%. Two decades ago it was 43% (humans are estimated at 4%).
Assistance
This is where intelligence helps you fulfill a task and improve your productivity. It can help you re-actively or proactively without asking e.g. searching for and making a booking. It can also include accessibility features like closed captioning on a video call.
Many products and services can assist you without necessarily being intelligent but it becomes much more powerful if the service understands intent with minimal effort on the user’s part.
Complexity
Intelligence can also be making sense of a complex set of information and processes, reducing it down to a simple form for the user and saving effort or time. It could be as basic as summarising large documents into brief descriptions or arranging a complex set of tasks and communication
Complexity could be an extension of Assistance but in this case it is applied to narrow tasks that require specific, deep knowledge.
Creativity
This is the ability to come up with new ideas through the mental process of connecting existing concept and knowledge into a new situation and coming up with better solutions. Creative people usually have more experiences and think deeply about a given experience. In the same way, computers could follow the same principles to exhibit creative skills successfully. We are starting to see experiments in creative AI involving script writing, music composition and art. In fact you can now hear the first complete pop song composed by AI.
Emotional Intelligence
One of the most difficult forms of intelligence to define and measure, it represents an understanding of subtle application of intelligence to achieve an outcome. This is a big challenge that AI faces as it tries to teach computers a very human attribute (which we can often struggle with ourselves). We can detect emotion right now at a base level i.e. happy/sad but emotional intelligence requires detection of a number of cues (which is more than just facial expression) and applying the most appropriate response.
Self-Awareness
This is the ability for computers to be self-aware, conscious, and able to act of their own volition. Be Afraid.
What does it mean to have a Super Power?
A user’s perception of intelligence has an impact, but more important is its application i.e. just adding intelligence on its own won’t always lead to long term engagement. Many ‘intelligent’ products we see in the market, although impressive, suffer from relatively low usage. Intelligence needs to be applied in a way that it is valuable for a user and improves their experience so that it becomes something they can’t live without. We need to ask ourselves if it involves significant improvement i.e. a ‘superpower’ or if it’s just a little bit better.
· Does it make a user achieve something significantly quicker?
· Does it remove friction compared to alternative options?
· Does it reduce the amount of effort for the user i.e. is it convenient?
· Does it give ‘superhuman’ powers to the user e.g. the best output is calculated using a million possible combinations that would be impossible to do any other way?
· Does it feel ‘magic’?
· Does it provide a superior experience or outstanding results i.e. it’s better than anything else?
Amazon’s Echo device combined with the artificial intelligence of Alexa, are a great example where the application of intelligence in a convenient way overrides the accuracy of intelligence, by making the user feel they have a superpower. Other voice based assistants existed on mobile but it’s Amazon’s innovation of an always on, home device which provided a magical user experience. Now you could talk to the device from anywhere in a room, whenever you wanted by simply saying ‘Alexa’.
With this in mind, an intelligent product today needs assistance, understanding and knowledge that must result in a superpower benefit for the user. All these factors need to be present in the product for success.
The Industry
Bots and Personal Assistants are a natural manifestation of where we are in the intelligence timeline. The factor that’s missing is providing real superpowers. Products will only survive if they accommodate user needs and those users will expect real, superpower benefits in intelligent products. Today we are seeing an up rise in products like Amazon Alexa, Google Assistant, Apple Siri, Microsoft Cortana, Viv, IBM Watson. Soon we will see self driving cars and specialist assistants with deep knowledge. The industry is undeniably applying intelligence and now we just need to make sure it gives the superpower abilities users will expect.
Please note these are my own personal views, and not necessarily those of my employer.