Hooking up a computer to the human brain or somehow connecting the brain to a computer is nothing new in the realm of sci fi literature. But say you want to merge the human brain and a computer together (think basic chip implant) and you do it successfully.
In computer science there exists a division of problems (P-NP) where computers can come solve a problem using an algorithm.
- Some problems can be solved in a reasonable amount of time with a deterministic approach (P) in polynomial time.
- Others can't be solved within a reasonable amount of time (NP-Hard) but can be verified in polynomial time if given information about the answer or information surrounding the answer (obviously oversimplifying a lot to a very great deal but this is the general concept for framing).
I am ignoring the differences between NP-Hard and NP-Complete for question simplicity, but if you know the differences feel free to use that to influence your answer.
The human brain is more than capable of predicting possible future outcomes based on incoming information, and short circuiting decisions based on patterns and behaviors.
Food for Thought
- When working together, what types of problems does the human brain help computers solve better?
- Essentially, could the brain help or enhance a computers range of solvable problems, especially for real time/dynamic problems (think military/racing/surgery/ language processing etc). If so, how would it aid a computer?
In this case I'm more concerned about the conceptual mechanisms than the actual how.
- Would it help in solving short circuiting based on past behavior/memories, race conditions, deadlocks, thread pool starvation, access/index out of bounds, preventing crashes or segfaults, program recovery from bad exits or inputs, prediction and confidence (confidence in a decision/the human version of the Viterbi algorithm) etc.
- Say someone is driving down a road and they see pebbles rolling down a hill in a very familiar way. Maybe the last time they fell like that, it was the warning sign for a landslide that was about to occur later. The driver would take that past memory and immediately use that to calculate possible outcomes while he is driving. The risks of a landslide, where he is, how likely it is to occur in his path etc. While a computer could also calculate this probability using a combination of algorithms and then do analysis to calculate the confidence of the outcome it created; a computer would need more than just sheer image processing.
- Someone weighing whether or not they should continue a conversation to change someone's mind when their target is saying questionable things but hasn't gone fully off the deep end yet. Sure, one could use natural language processing to break down the sentence, but that brings in a whole host of issues surrounding cultural usage of words and phrases versus their textbook meaning.
- Interpreting sentences differently from how humans would because certain words can have multiple cultural contextual meanings or even grammatical meanings (e.g. A computer can hear "She saw the ant" and interpret it as "She saw the ant" or "She saw the aunt". Casual swearing between friends is another example.) But a person listening to the conversation would be able to immediately pick up on the context (I am aware that humans are capable of misunderstanding each other, but I'm choosing to ignore that subset of problems for now since such a situation can't even be verified for correctness by a human let alone an NLP classifier)
The Question to be answered by you
How would the human brain (mechanically) aid a computer in problems previously outside of its solvable range in real time/dynamic problems, given a hybrid between human and machine was possible? Think mechanisms here, not scenarios.