As with others, I would suggest that you consider finding someone who you can collaborate with. That person should be doing research in AI as well. This serves two purposes;
1) It will help you determine if you can explain what you're doing to someone else, well enough that they can duplicate your results. If you cannot do this, then writing a paper will be a failure for you.
2) If they are in academia, then you will not only have someone who already has an understanding of how to get published, they can help with the paper's structure to make it useful for others (increasing the likelihood it will get published).
The challenge is not that you are a "college drop out", the challenge is that you didn't get a chance to learn all of the meta information that college was going to teach you about how to structure a research paper, how to explain the problem and your solution, how to help the reader understand the "big question" you are trying to answer, and the effort you have gone through in making sure you haven't deluded yourself your on to something new (that you have avoided confirmation bias).
1) It will help you determine if you can explain what you're doing to someone else, well enough that they can duplicate your results. If you cannot do this, then writing a paper will be a failure for you.
2) If they are in academia, then you will not only have someone who already has an understanding of how to get published, they can help with the paper's structure to make it useful for others (increasing the likelihood it will get published).
The challenge is not that you are a "college drop out", the challenge is that you didn't get a chance to learn all of the meta information that college was going to teach you about how to structure a research paper, how to explain the problem and your solution, how to help the reader understand the "big question" you are trying to answer, and the effort you have gone through in making sure you haven't deluded yourself your on to something new (that you have avoided confirmation bias).