Correct. I was just putting out there that not everyone thinks AI is going to destroy the planet tomorrow. The fear mongering articles have about as much meaning as the geologists that say California can have a massive earthquake in the next 30 years. I don't need a P.H.d to come to that conclusion. Same logic applies here.
How do we define intelligence? The type of intelligence needed to take over the human race or make us obsolete is a type of intelligence that I don't have to worry about in my lifetime. In fact, I think it's so far off we have a better chance of colonizing another planet. Now making smart tools to handle everyday things is surely attainable and will drive a new economy. But a conscious technology that will make it's own decisions and act in its self interest I think not.
I have fiddled (Not seriously) with neuro networks and computer vision. Far far far from any type of AI scientist/engineer but I think it's cool stuff and so on occasion in my down time I fiddle. There are API's that pick out patterns. The system has no idea what the data represents. It just knows the statistical side of certain points of a photograph. For example, my family all have the same eye/bridge of nose shape. The facial recognition frequently gets my daughters mixed up. The intelligence is not really AI, it's smart technology that uses statistics recognizing patterns but it is not smart enough to cognitively decide - "hey what other element can I look for to match these two pictures?" A programmer can add the code to say, "hey if user corrects more than three times then scan the rest of the photograph and look for other marks like freckles, which are defined by function F and birth marks, which are defined by function B". The self awareness, consciousness stuff is what is lacking and really I can't see ever coming to fruition.
Incidentally, neuronetworks are really cool but are slow to learn. This is one of the reasons they lost popularity in the AI circle until recently as graphics chips have proven to be faster with computations. But even with that they are slow and require assigning weights to inputs. It gets complicated and big real fast. Every permutation of every input needs to be analyzed and weighted depending on how deep the network is it could take a long time to solve even the simplest of problems.
Another area that is lacking is storage. Where is an independent machine going to store all the data needed to make conscious decisions? Storage technology IMHO, has not been able to keep up with what's needed by super smart machines. I spoke to a Manufacturing engineer on one of my flights who works with Micron and he was telling me all about some of the advancements they are making in storage. It's pretty cool stuff but there are physical limitations that we need to overcome to put more bits in smaller areas. Then there' the issue of powering the cybernetic beasts. We've got a long way to go before we start fearing our smart technology. Capitalize on it? Absolutely. New economy? Absolutely. SkyNet? Nope. I think to get to that point we will need to use DNA based organisms like what these people did:
Artificial Intelligence Created Using Human DNA
Those are my .02 and I'm sticking to it.
