Do you travel to work in a horse and cart?

It is now almost a year since the Go computer program AlphaGo, constructed by DeepMind, beat the world Go champion, Lee Sedol .

As I understand it, a key success factor for DeepMind was that Go is a rule-based game; and machine learning from DeepMind ‘loves’ rule-driven analyses and being fed thousands of historic completed games from which to learn.

I have pointed out that geology is a rules-driven science and that machine learning could be applied to seismic (perhaps especially 3D seismic) interpretation, there being thousands of completed interpretations available for ‘learning’.

Perhaps this is a half-baked idea? I do not yet see any evidence of such disruptive transformation of seismic interpretation.

Approaching this with an inappropriate sense of humour, I suggest this leaves us in the position of ‘Art can only be done with paint brushes’ or ‘Presentations can only be done with transparencies and an OHP”!*

Perhaps this is the best analogue of the serried ranks of seismic interpreters working in our industry…..

and inclined to travel to work in one of these:

Putting  my serious hat back on, it seems to me that the current combination of humans and (interpretation) workstations is both too high cost, in an era of lower oil and gas prices, and not as effective as we need, given that there are plenty of examples of mis-interpretation – incorrect correlations, incorrect chronostratigraphy, implausible structural geology, unlikely reservoir distributions…….

Our report card would read: “Must do better! Can do better!!”

* to the first person who can 100% prove to me that they made a presentation using transparencies and an OHP in 2016, I will award a prize of a lino-cut picture, based on a scene from the petroleum industry!