Daniel Lemire's blog

, 1 min read

Stock-Picking Computers

In my we-do-not-have-strong-AI-yet category, I found this juicy quote on slashdot:

Investment firms fall over themselves advertising their latest, most esoteric systems,’ said Mr. Lo of M.I.T., who was asked by a $20 billion pension fund to design a neural network. He declined after discovering the investors had no real idea how such networks work.’ Why did he decline? ‘I’ll build a stock market neural network for a couple thousand (you’ll just need five thousand years of market data to train it before it actually works).

I was approached myself to work with a company that did stock picking by computer. When I suggested they look at having the computer help the human being make better decisions, they dismissed the whole notion as being silly. Why help the human being when the machine can take over? Why have so little ambition?

Well, the truth is, now wait for it, computers are less intelligent than human beings! I can’t believe I need to write down something like this, but it is fortunate that there are still Machine Learning researchers (at MIT) who do not fall in love with their topic to the point of living under the illusion that it is easy to mimick human intelligence.

Now, you’ll excuse me, I have my daily batch of spam-that-got-through-12-layers-of-machine-learning-spam-filters to delete.

Source: Slashdot.