I understand you're the sole developer of the project and that you need to get help in any way you can.
But there are some really basic problems.
Many have already commented about the continuous reload thing. It can be fixed, apparently, by closing the tab and reopening. Apparently. That's what I did and it worked. But it's a big enough problem that needs a deeper look. That's why someone specifically trained to test would help you much better.
There seem to also be a bunch of bad-loading problems. Sometimes stuff doesn't load. This is so bad that you even have a message asking the user to reload or re-type the query until it works. Yes, I understand about the money restrictions, but throwing that against a whole community will hardly be productive.
I mentioned "bad-loading" and not simply not loading because sometimes the wrong cover loads for a certain book in the results and you hover over it and it's a different book, click it, and it gets the correct cover.
Again, I understand that you want help and are alone. But these problems are big enough to be showstoppers. In the sense that it will be much harder to get anyone interested if thing don't even load correctly or at all, or reload infinitely.
Now...
The mind-map seems like a nice feature. But it lacks links between the bubbles (at least in my browser) which feels weird. It makes it look more like a bunch of bubbles randomly thrown in there than an actual mind-map with a structure or relationship between nodes.
Also, it frequently features duplication. Just use the given example of The Great Gatsby and you'll get a second "The Great Gatsby" bubble as if it was a different book. I tried randomly "Neuromancer" and there were 2 extra Neuromancer bubbles. It doesn't look like they are duplicate entries in the dB but more like a problem building the mind-map.
I also get some random bubbles missing their title.
As for the "Book-genie" recommendations I can say very little. I tried the "Inspiring books with aliens" example and I got "Seven Brief Lessons on Physics" along Cixin Liu. Are these results good? I really can't say, but I don't feel like a physics book matches the mood. But in any case, one would have to test this much more deeply to actually evaluate if the recommendations are any good, or even simply any better than a keyword based search.
There's also more general design and UX comments that I could say. But again these are hardly specific details. It feels like you'd need someone to dedicate some time exclusively to that part and re-do things in general.
----
In any case, I do hope you get the help you need to make this work. I just don't think putting this up in its current state against a large audience will do much good.