When computers can reason, will they want us around?


It is so encouraging to see maintream press, like the Financial Times, reporting on what I think will become the the key issue of our lifetime – conscious machines.   Although this article pretty much dodges the most intriguing aspects of the debate over AI, rational computers, and consciousness  it does offer some insights into the state of the science in the semantic web, where AI routines are used to create a better search experience.

One researcher suggests that he’s given up on the idea that simply creating a massive neural network and priming it with some info will lead to conscious thought but I still think that hypothesis has not been tested nearly enough because our computing capacity is still far short of what you and I have between our ears in the three pounder we call a brain.    Brains offer a spectacular number of individual neurons, and in turn a simply staggering number of interconnections between those neurons.   It will be another decade or so before we have that processing capacity in computers, but it will certainly happen.   I’ll be surprised if our consciousness and intellectual abilities are as profoundly amazing as we like to …. think they are.    In fact I’d wildly predict that we’ll have conscious machines within 20 years and that those conscious machines will surpass us in every imaginable intellectual and creative ability within months – probably days – of consciousness.    Is this because I’m hugely optimistic about technology?    NO!   It’s because I’m hugely confident we overrate our feeble human abilities, which I’d suggest are just a few shades richer than those of our dogs and cats.

Compete.com sale a champagne moment? Not at ~8% per year return it’s not.


Update:  Silicon Valley Insider is reporting that there is an additional 75MM in the deal as an “earnout” over the next several years.   That may make this deal sweeter than it appears at first.

At first glance you’d think the sale of website COMPETE.com, which measures web traffic, for 75MM must have been a big payday for a lot of folks.   However as Venture Beat notes some 43MM of venture capital had been poured into  COMPETE over the past 8 years.  

Assuming most of this came at the beginning of the cycle, and assuming most of the 75MM is going to the VCs, the return on this VC investment would be a very modest 8-10%.    If the founders and workers also had a decent stake in the sale this return could be lower – approaching what the VCs might have realized with long term CDs over the same time period.     Break out the champagne?

I’ve noted before the dirty little secret of many “successful” venture capital deals – they often make a very modest return when time is factored in properly.   In fact it appears that *most* VC deals lose money for the players.    Data is sketchy, and obviously only the winners are happy to share the details making it very difficult to analyze this since many (most) of these deals are not in the public record.  

Sure there are VC winners like Fred Wilson and Jeff Clavier, both very clever VCs who blog some of the details of their failures and successes.    However I think this is not typical, as Jeff even suggested here at the blog some time ago.