AI is now changing how we are going to use many services in the future. Voice and text are becoming more important for many services. At the same time, there must be enough data for services and AI training. That means devices that use AI must be connected. Does this mean we are finally starting to see the emergence of ubiquitous computing, truly connected smart devices and totally new ways to use them?
Wikipedia summarizes ubiquitous computing as a concept in software engineering, hardware engineering and computer science where computing is available anytime and everywhere. In contrast to desktop computing, ubiquitous computing can occur using any device, in any location, and in any format.
Meanwhile, ‘ambient computing’ is a broad term used to describe a setting where smart devices, AI, machine learning (ML), data, and human activity all work together to produce a result without the need for human command or interaction.
Both terms have been in discussions for years. However, this has also diluted their value, as there have been expectations that they would become real long before now.
Anyway, the reality is that we haven’t seen the IoT revolution as we originally imagined it. But at the same time, connected devices, sensors and data utilization are developing fast. It just might be that things are happening in another way than the business plans of ten, five or even two years ago expected.
Four layers needed
When we think about AI services today, they basically need four layers:
- A device near the user that provides the user interface
- A service that can execute tasks
- The data needed to execute those tasks
- Data to train the model.
Currently with generative AI, we basically see models where the user interface is either a phone app or an internet browser on a phone or desktop (although services like Siri and Alexa can use different devices for the UI). The actual AI service is in a cloud somewhere, as is the data it uses. But we are in a very early phase of this development right now.
I have written earlier about connected appliances still being unconnected, sensorization and poor usability of services in cars. All of these are also relevant to new AI services in that it involves devices that collect data, and these devices should utilize their own and external data and offer a good user experience for the services they use.
Making microwaves and washing machines smart (i.e. usable)
To get an idea of how AI could really make a difference, think about the complexity of appliance UIs. I have a German microwave and a French washing machine. Using them requires pressing several buttons, turning several knobs, and sometimes checking the manual to see what all the different options on the display actually mean, as they can be quite cryptic.
Why can’t I just say, “cook my normal porridge,” or “do the normal wash for mixed laundry at 40º Celsius”? Or perhaps the appliances could propose those things when, say, I put something in the microwave before 7am or something in the washing machine on Sunday afternoon.
And when we start to talk about real AI, we can go much further. An oven should really know the optimal cooking procedure for different things. The laundry machine should determine the best program for whatever I put in it. Then we can return to the classic vision we had 30 years ago where your fridge sends an order to your grocery store when you’re almost out of milk.
The components for all of this are already starting to be available. But questions remain about how this is actually going to happen:
- Who collects and uses your data from your devices and your home (where are they)?
- How much intelligence do different devices need? Should they just be dumb terminals that take instructions from us and AI?
- What is the network infrastructure going to look like? Will we really see edge computing in some formats, or do we stay with more centralized clouds?
- Is the phone going to be the main device, or do we shift to more headless services?
- How do we develop all this so that users can finally have control over their data and get more value from it, instead of the current mess where we have all kinds of devices that try to do something smart, but don’t really serve the users?
Some of these questions are very old – e.g. how much intelligence and processing should reside on the device vs elsewhere in the network. We have seen this with mainframe vs personal computers, smartphone apps vs browser-based services and local data storage and servers vs clouds. Perhaps there is no final solution for this, but rather it is more like a pendulum.
Winners in the era of AI-powered smart devices
It has been said that most services in the future could utilize AI in quite similar ways. So, who will be the winners in this era? The most common answer is: whoever can utilize the best and most relevant data for their services, offer the best user experience, allow customers to utilize their own data whilst also respecting their privacy.
Obviously we cannot yet know what will be the final model for using all connected smart devices with AI. But this is definitely the time where we have a lot of opportunities to make new better devices, services and businesses.
It is possible that this development will change many things. We will have solutions to manage our private data and utilize it how we choose. We will have a totally new user interface for cars and home appliances, and we will have many new sensors and microchips everywhere to enable totally new services.
If nothing else, it’s interesting to think that AI may make our microwaves and washing machines finally easier to use.