When we think of automation, we invariably think of autonomous driving. Due to its complexity, however, we classify autonomous driving by a range of distinct levels. For example, we have to consider driver intervention and attentiveness levels required by the driver. Inspired by these different levels of automation, we decided to ask ourselves how it might affect our daily work in a design consulting agency.
At designaffairs, we love engaging with others and taking part in face-to-face meetings.
That’s when we amplify our creativity, create and share our ideas, debate and discuss our hypotheses and strive for the best solutions. It’s part of our identity and it’s what makes us like a family.
As an exercise in automation, let’s take a look at the apparently simple task of scheduling a briefing meeting with your project team. What is the current level of automation in this task and could it ever be fully automated? Furthermore, which steps are necessary to achieve more sophisticated automation? Are the five levels of automation applicable to meeting scheduling? What might change for us when automation takes over fully?
Level 1 — Assisted:
Assisted scheduling is what most of us experience when we use tools like Outlook, iCal or Google Calendar. The meeting organizer adds the participants, decides on a meeting time and checks for an available room. Conflicts are highlighted and more suitable times are proposed. Nevertheless, in the end, the organizer is responsible for actively setting up the meeting, using all the information and proposals provided by the system. The user is fully in charge and has full control over scheduling.
Level 2 — Partly Automated:
For meeting planning, partly automated means that the system is already able to schedule a meeting. First, the user needs to provide information about the length of time required, the meeting participants and the room it should take place in. Based on this information, the system can then provide the most suitable option for the meeting. All the organizer has to do then is to decide what to do with the proposed meeting: accept, reject or edit. This means the user still is in control, although the system is doing the majority of the work.
Level 3 — Highly Automated:
The system still needs basic input from a human, as with levels one and two. However, at this stage, the system not only provides a suitable meeting option but is also able to reschedule your meetings and those of the other participants, if required. In order to be able to do so, meetings need another property — priority. The key point of level three is that the system can now take action to manage schedules. The only remaining user interaction is to actively confirm any rescheduling made by the system, with a further possibility to reschedule manually or input optimizations.
I have this
Level 4 — Fully Automated:
The system automatically schedules a meeting with minimum input provided by the user, such as “I need a briefing meeting with the team”. Time slot, duration, meeting participants and even content will be managed by the system. To do this, it will be connected to a wide range of data sources that allow for optimal scheduling. It will check your usual presence in the office to find out which slots suit you best. It may even consider the estimated wellbeing of the meeting participants to ensure everyone is at their best mental level for the meeting. As an employee and user of the system, that also means you have to disclose information about yourself. Lastly, users will only be informed of changes in their calendar; there is no need for confirmation. The system is taking over control and responsibility.
Level 5 — Autonomous Scheduling:
The human component is obsolete. The system, in this case a powerful AI, will automatically schedule a meeting with all components. Since the need for any user action has been eliminated, the initial kick-off for scheduling a meeting is not provided by a human anymore and might instead be initialized by other data, e.g. receiving confirmation of a new project. Being an interconnected, fast learning and highly intelligent AI, the system will know when, where and how often a meeting should be scheduled. At this level, the user has handed over control completely. The human input is completely passive; the user is merely a meeting participant being managed by the system. In this scenario, it is essential to completely trust that the system will make the right decisions, considering all the important parameters such as the participants’ behaviors, preferences and skills.
We began this exercise by attempting to define the key points of a scheduled meeting, using automated driving as a model for our conceptual build. For classification purposes, we identified four main attributes:
- Control of the process: The higher the level of automation, the more control is shifted from the human to the system. Higher automation means less human control of the work schedule. With this loss of control comes less responsibility and more time to focus on other tasks.
- User Input: From level one to four, the user always has to initiate the meeting request. However, with advancing automation, the system needs less detailed user input. Whereas the time, duration, location and participants all have to be provided in level one, by level 4 the system chooses these properties itself. The autonomous meeting in level 5 is scheduled without any user intent at all.
- Data sharing: For the system to work properly, it needs all sorts of information that can only be considered if the user gives permission to evaluate it. Therefore, the system might need to know of private meetings that cannot be rescheduled easily, e.g. a doctor’s appointment. It needs to know your preferences regarding time slots and to know the time of the day you consider yourself to be at your most productive (especially important for a creative workshop, for example). This implies that the user is monitored and conceded most of his or her self-determination regarding prioritization.
- Trust in the system: In order to accept the automation, the user must have high very confidence in the system. In effect, they must trust it completely (level 5). The user must be sure that the system considers all of the same aspects that an individual would have in mind when planning a meeting, since there is no possibility to intervene. In short, the system must be totally reliable, making the right decisions every time.
With this experimental excursion, we are preparing ourselves and our clients for an AI-driven future; to open our minds to what might come but also to reflect on how deeply technology is already ingrained in our lives.
Ask yourself this: how did you feel when you first experienced an automated system? Was it exciting or scary, helpful or disturbing? Looking to the future, to what extent will technology influence, manage or even control our lives? Why aren’t we there yet? What keeps us from already living in a fully automated society? The technology is here and the data could be collected to make it happen.
It might be due to a lack of trust in the system or the uncomfortable feeling of losing control and having to rely on a machine. But automation is already happening. It happens when your phone offers you news based on your interests, when it plays music based on your preferences, when it suggests an alternative route based on traffic information, and when it provides available meeting rooms. Automation gradually sneaked into our lives and maybe that’s the way we need to adapt to it to embrace its benefits in the future.
This quick exercise is a glimpse of what the future might hold. Certainly, there’s a lot more research that can be done and future scenarios to be explored seem almost endless. What we have discovered to date, however, is that a shift away from responsibility and control is absolutely needed to make an autonomous system fully functional and effective.