When I mentioned a truck engine, I meant the fuel used to drive the truck - it has energy in the cold, motionless gasoline in the tank. It has to extract the energy from the fuel, like a nuclear rod. Data/knowledge extraction is same, you extract free insights using the top n predictions.
Training a predictor to predict truck driving....as I said, you need directions, and associated motors to carry them out. Doing this would be more like a user asking their driverless truck how to get to destination X, then it predicts the Answer. This is just a text predictor, (well if you use gas pedal, dashboard, GPS, etc as context, it is just the advanced version of a text predictor; vision etc used), like the ones in the Hutter Prize, so we are already making a truck driver predictor, except the ones in the Hutter Prize won't know how to get me from A to B as the dataset lacks street names (I think). So, a text/vision predictor with big diverse data can teach us the truck driving directions but someone needs to follow the route if AGI only speaks to us. This makes sense because the AGI can, using text, know the words press, pedal, move, drives, left, forward, right, return, city, 'streetnames' and which follow each other. So text has many patterns and free insights in itself connected to the real world, but we'd need to follow its instructions if it didn't have a body. So ya, a text predictor 'AGI' can drive a truck, if we follow instructions. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Tc67faac3048278cf-M8e571b4cbf97e71bd59dca24 Delivery options: https://agi.topicbox.com/groups/agi/subscription