Although, yes, computers may produce flawed outputs based on flawed inputs, I don't think that their lack of ability to live up to the expectation of perfection that people have of them should be a reason to exclude them completely from politics. Seen another way, computers may - in some cases - have different points-of-view and different insights than existing participants. Finding a way to gradually introduce their views in a way that demonstrates their strengths (and demonstrates the benefits of utilising those strengths, in terms of improved legislation) could be beneficial.