By Jas Mahrra, on 10 July 2015
AI can and has been oversold, over packaged, over expensive & doesn’t do the job. That’s been a recurring pattern since cybernetics morphed into AI after WW II. AI investment waves have come and gone. As the recipient of a recent PhD in the field put it, today’s machine learning is “statistics on steroids”. Yes, that’ll doubtless provide a financial engineering competitive advantage for a narrow window of time but we’re some way from the singularity and each competitive advantage snatched at and lost probably doesn’t amount to a singularity let alone the singularity that those suffering narrow inductive rational rapture have long fantasised. Uncertainty still wins through as the Flash Crash perhaps demonstrated.