Trust in AI and Robotics
Sanctuary is working towards human-like intelligence in general-purpose robots that will help us all work more safely, efficiently, and sustainably while also creating new opportunities and jobs.
Sanctuary is working towards human-like intelligence in general-purpose robots that will help us all work more safely, efficiently, and sustainably while also creating new opportunities and jobs.
Accepting general-purpose robots as digital colleagues will be about trust: trust in one another is essential for a successful and productive work environment.
Engineering trust into a platform to deliver the right customer experience and user experience is about architecting the right 10 Dimensions of Trust, and delivering digital colleagues that demonstrate those 10 dimensions.
1. Colleagues That are There For You
Think of great colleagues you’ve worked with; they’re there for you, they respond when called upon and, if they’ve got a problem, they make sure someone is available to cover for them. As with any digital service, website, or application that you’ve come to rely upon, high availability and reliability is critical to earning trust.
High availability and reliability, enabled with hot- and cold-standby backup capabilities is architected into the Sanctuary solutions and general-purpose robots with highly available connected fleet management services.
2. They Remain Focused and Learn Quickly
As managers, when we delegate a task, we want our colleagues to remain focused and avoid getting distracted. It is also common understanding that learning a task is faster when one focuses, rapidly learns from mistakes, and adapts to feedback.
Sanctuary general-purpose robots are engineered to be able to accept feedback, update rapidly and, when appropriate, update over-the-air (OTA). As well as using advanced machine-learning and cognitive reasoning, Sanctuary’s fleet of general-purpose robots is engineered using DevOps and OTA technologies to manage and maintain the fleet with up-to-date knowledge and skills. As an enterprise adapts and continues to evolve new processes, it can trust that Sanctuary general-purpose robots are up-to-date.
3. They Learn from Mistakes
Continuous improvement, Lean, Six Sigma, Agile engineering, and digital transformation are some of the approaches that enterprises have adopted to turn ‘mistakes’ into ‘lessons learned’.
We all try to learn from our mistakes and the mistakes of others, and these techniques are engineered into Sanctuary technology–learning from mistakes is part of the cognitive platform. But more broadly, Sanctuary general-purpose robots and fleet management capabilities are designed to allow individual robots deployed at a customer site to adapt and learn skills developed with other robots at that customer's facilities.
4. They Think Like You Do
Working with others is so much easier when they understand your preferences, and present information in a way that is familiar to you. As well as learning skills and techniques, Sanctuary robots are engineered to learn the preferences of their colleagues, easing the introduction and adoption into the workplace–so you can work comfortably together and in sync. Your digital colleague can present information and skills the way you want it to be presented.
5. Trust but Verify
There’s a principle in management to ‘trust but verify’. When you delegate a task to a colleague, you trust they’ll do it, but verify that it’s been done correctly. The concept of ‘trust but verify’ in digital systems is similar: delegate the task but then audit and review to ensure it has been done correctly.
One of the key principles of Explainable AI is ensuring that the system can explain how it reached conclusions. Sanctuary’s cognitive platform is designed to allow operators and managers to audit the decisions, track the individual activities within a task, and understand that if a robot choice needs to be corrected, then it is clear why it chose to do the task in a particular way.
When you ask a co-worker for help with a familiar task, how often have you dealt with the situation where a decision made one day is different to the decision made previously? How often have you asked, “What changed?”
Being able to audit the logs and track through the decision making process, data and system updates, coupled with the traceability in the Sanctuary cognitive platform, makes it easier to diagnose and correct behavior if needed.
6. Safe to Work With
Sanctuary takes safety extremely seriously; we are engineering in a safe environment to build safe robots. The Sanctuary general-purpose robots are designed and engineered to minimize the risk to people, such as the physical form factor, power systems, actuation and locomotion. The software and hardware is engineered with human interaction and safety first.
7. Demonstrates Common Sense
Interaction between people is built upon ‘common sense’ and ‘common knowledge’. Those small details about the way the world works is built upon years of experience of common social knowledge. When teaching a colleague about a task or job, it would be very frustrating if you had to explain basic knowledge that is obvious to us all. Historically, common sense has been something that has been missing from computers, artificial intelligence, and industrial robots.
At Sanctuary, we’re engineering our cognitive platform and general-purpose robots to encapsulate common knowledge and an ability to reason about that to display a level of common sense.
8. Maintains Privacy and Security
Often in our day-to-day activities, our personal and private information and knowledge is shared with our colleagues, such as critical information about our identity and personal preferences. We grow to trust those colleagues who keep that information to themselves, and only share it when we’ve given permission. The same has to hold for our digital colleagues.
At Sanctuary, we are engineering our general-purpose robots and the cognitive platform to encapsulate up-to-date understanding and techniques for preserving privacy and confidentiality of our customers, and their customers' data. In addition to ensuring that our cognitive platform aligns with the appropriate and applicable privacy and security standards, our platform is built to be rapidly updated, helping to keep us aligned as these privacy and security standards evolve over time.
9. Aware of Bias and Data Drift
We all work to address bias in our decision making, and be aware of our own bias and to address it. In recent years, the industry has also become aware of the risk of bias emerging in AI systems–and that bias emerges from the data that is used to train machine-learning algorithms.
In adopting industry standard techniques for minimizing risk of bias, at Sanctuary we work hard to ensure our cognitive platform is designed to minimize the risk of bias by monitoring and reviewing data we use to determine if there is a drift or change in any inherent bias.
10. Operates Legally
We diligently abide by the best software and hardware practices, including those specific to our customers’ industries, and fit within national legal guidelines.
Working with trusted technology that performs reliably and accurately will have a tremendous impact on the entire work environment. That trust will enable people to focus on more important tasks and drastically improve their own efficiency and workflow. A digital colleague who you can trust will create the opportunity for you to focus on what you do best and delegate the rest to them.
Matt Ellis
SVP Engineering
Sanctuary
Sanctuary is on a mission to create the world’s first human-like intelligence in general-purpose robots that will help us work more safely, efficiently, and sustainably, helping to address the labor challenges facing many organizations today. Sanctuary's growing list of customers and investors represents a wide variety of industries across Canada, the U.S., and other countries around the world.