Our Approach For Building Digital Brains
Our development methodology at Mindsmiths ensures that AI products can be built even in situations where there is not enough relevant data to train machine learning algorithms. Just like knowledge has been passed down for generations from one human to another, we make it possible to train AI algorithms on synthesized human expert knowledge.
First we learn from human experts, then we enrich it with relevant data. Through a process of knowledge discovery, a prototype is built based on the knowledge of an expert in the field. Then we push the prototype into the production environment and start collecting relevant data. The newly acquired data is then being fed into machine learning algorithms to update models. This learning loop never stops improving models.
A core loop is a term from game design but also an essential component of building apps that users love and return to over and over again. A core loop is the primary flow of user experience - a set of actions that are cyclically repeated while using the app. If well designed, a core loop induces a sense of achievement and engagement within the intended group of people.
To develop an AI brain that displays human-like compassion, it’s important to have a perfect understanding of the human users and what makes them tick. To achieve the perception of emotional intelligence displayed by our digital brains, we go deep in our quest to discover knowledge.
Our knowledge discovery experts immerse themselves fully into the world of the user, employing various techniques to discover as much knowledge as possible. Even if it means going through hell and back, we want to get the best possible understanding of the users' fears, motivations and the different emotional states surrounding their specific situation.
Our toolbox consists of techniques borrowed from many disciplines, like structured and unstructured interviews, undercover investigations, thought experiments, video documenting and ‘the lab rat approach’.
Our aim is to democratize expert knowledge and make it available to everyone, so that even the most underprivileged have 24/7 access to medical or financial experts. The most demanding part of our work is always building a copy of expert knowledge in the form of a central, digital brain.
What we want is a digital brain capable of making decisions that create value for the user without doing any harm. Once the brain is developed and tested, it’s ready to be integrated into an app, a website, or chatbot.
When we develop a dedicated digital brain with expertise in providing healthcare for chronic patients, for example, then this brain can be used in thousands of healthcare apps and institutions, with only minor work on finishing touches. The cost of development is therefore spread across multiple vendors and the brain becomes highly accessible to the end user.
As we succeed in building an AI solution that can help one person, that superpower can then easily be scaled to help an unlimited number of people.
Human-in-the-loop (HITL) is a branch of artificial intelligence that leverages both human and machine intelligence to create machine learning models. In a traditional human-in-the-loop approach, people are involved in a virtuous circle where they train, tune, and test a particular algorithm. HITL intervenes when the algorithm is in the prototype stage or is unable to solve a problem. Mindsmiths employs humans in the loop to ensure our AI is trustworthy and always accountable.