We can define free will however we choose. The problem isn’t with how we define it, but rather with the consequences of any particular definition.

For example, if we define free will as “the ability to choose”, I would say that’s a reasonable definition, since we do indeed have the ability to choose, and choice seems to be involved in most definitions of free will. But what are the consequences of this definition? Anything that has the ability to choose is now defined to have free will: computers, toasters, thermostats, plants, etc. In fact, depending on how we define choice, we might even be able to say, with this definition of free will, that everything has free will, and it then becomes apparent that the definition isn’t really defining “free” or “will” in any usual sense.

Having said that, I would agree that relating free will to our inability to predict the future seems reasonable. But the consequences of such a definition will force us to admit machines and computers and flowers into the “free will” club.

Our inability to untangle the complexity of the decision-making process and predict its outcomes contributes significantly to the illusion of free will (maybe it’s even the source of the illusion?), but is not itself free will.

Among those who believe that free will exists, they would say that people have free will and machines don’t. All of the proposed definitions of free will that I’ve seen so far are either too inclusive: they allow for toasters and plants to have free will, or arbitrarily exclusive: the definition arbitrarily excludes non-human entities and objects (for example, defining free will in terms of whether or not an entity has a prefrontal cortex!)

So when thinking about a definition of free will, ask yourself if it is overly inclusive or artificially and arbitrarily exclusive. If either is the case, then the definition is not adequate.