Contexts

Collaborative Intelligence (CINTEL) is relevant to many, but not all, contexts where machines are employed. We focus on contexts where a person working with a machine can get better results than either can alone – hence we are better together.  

Machines can work well on their own where the problem space, whether physical or digital, is clearly defined and specified. However, this is not the case for many of our most challenging problems. This is where collaboration is needed. It is a journey; much cannot be known or specified in advance. 

CINTEL is most applicable to complex tasks in which humans have a comparative advantage over machines (e.g., strategic thinking, ethical reasoning, or goal setting). The CINTEL Future Science Platform within CSIRO will focus on contexts that will benefit most from long-term collaboration between humans and machines. These have the following characteristics: 

• Complementarity: Humans and machines bring different, but complementary, aspects of intelligence relevant to the context. Together, they can perform better than either alone. It is important to recognise the strengths, and limitations, of each. 
• Meaningful relationships: We are not focussed on a simple division of labour (e.g., a system runs the model, a person writes the report). The greater opportunity here is designing systems for tasks where best results are achieved when people and machines work together.  

• Empowering: True collaboration benefits both parties, and, as Australia’s national science agency, we are focussed on developing systems which work for, rather than against, people.  
• Sustained partnerships: We will focus on contexts requiring ongoing collaboration between humans and machines, not situations where humans train machines to replace them (or others). In CINTEL, humans and machines will learn from each other. 
• Beyond definable: The problem, and the context in which it fits, cannot be fully specified. For example, while the high-level goal might be clear (e.g., have a sustainable farm), the process to get there may not be – this gets worked out collaboratively. Goals may also be fluid and not fully articulated (e.g., what is a sustainable farm anyway?).