Well - the market is a giant processing machine continually spitting out a stream of price data, which really are operating instructions for the economy.
If price of copper increases week after week, the market is telling the world we need more copper, and as the price of copper rises, more and more marginal mines become profitable and get re-opened by operators. Prospectors start looking for new mines. The world also starts to look for copper substitutes - can we make do with another alloy? And conservation to improve material efficiency - scientists start looking into ways to reduce copper usage. In this way, the price of copper is the operating instruction that tells the world how to mobilize it's resources accordingly.
Information theory tells us that if a stream of data is delivered in an inefficient manner, then there is some predictability in the stream of data. A perfectly efficient delivery system for information would appear to be completely random, without pattern.
Now, while computers might continue to get better and better at attacking patterns there are a few factors that won't go away. I'll try and name 3.
1) the cost of capital will ensure that the arbitrage opportunity will remain at a positive spread to the risk free rate. Thus the average spread on arbitraging between say cash SPX and the futures can never converge to zero because even if a computer could do it, the computer and its maintenance costs some capital, and that capital must earn at least the risk free rate + a premium or else we would unplug the machine, sell it on ebay and put the cash into T-bills.
2) mimicry is endemic to human nature. as infants we learn by mimicry. as adults, we see what others are doing and we might copy. as such, ideas in society tend to gain momentum - whether it's a trend, a meme, a sea-change in opinion. This creates positive feedback mechanisms in all things human. People come around to an idea at different speeds, thus creating momentum and arguably trends.
3) the statistical properties of price time series data are linked to human behaviour. our reaction to the pain of losses and the overconfidence of wins is deeply rooted in human nature and will find expression in how we behave in markets. Fear and greed will always colour price movement and the interaction between weak hands who sell in a panic and strong hands who wait for such a moment to take advantage of the weak has such primal roots and might explain the volatile ie regressive nature of bear markets.
