Here’s something I learned in the process:
1) Gemini Pro 1.5’s context window is far bigger than the next closest models. While Gemini Pro 1.5 is comfortably consuming entire works of rationalist doomer fanfiction, GPT-4 Turbo can only accept 128,000 tokens. This is about enough to accept Peter Singer’s comparatively slim 354-page volume Animal Liberation, one of the founding texts of the effective altruism movement.
2) Gemini Pro 1.5 can use the whole context window. In my testing, Gemini Pro 1.5 handled huge prompts wonderfully. It’s a big leap forward from current models, whose performance degrades significantly as prompts get bigger. Even though their context windows are smaller, they don’t perform well as prompts approach their size limits. They tend to forget what you said at the beginning of the prompt or miss key information located in the middle. This doesn’t happen with Gemini.
Source: Gemini Pro
The only other thing I found interesting was this:
I’ve been writing for a long time about how transformer models might become copilots for the mind—and end our need to organize our notes forever. Gemini Pro 1.5 is a step in that direction. For example, recently I was writing a piece about an effect I’ve noticed that I’m calling “I can do it myself” syndrome, where people tend to not use ChatGPT and similar tools because they feel like they can get the same task done more quickly, at better quality, if they do it themselves. It’s like inexperienced managers who micromanage their reports to the point of doing most of the work themselves, guaranteeing it’s done the way they want it to, but sacrificing a lot of leverage in the process.
Basically, this is equivalent of outsourcing your thinking to AI. The workflow as “reimagined” here is simple- highlight/notes and dump it into a large file. Then let Gemini run riot with the notes. Ask questions on your own notes without ability to recall. Your memory base has been completely eliminated.
This is as dumb as it sounds; the VC types are interesting people. They try to coax out the “exclusive features” from the PR material handed out to them. Then its a canned response from the company in question- “the team will review your suggestions”.
Ignore the recommendations but I linked to it because it offered a different perspective of Gemini (and from a first hand user). I understood tokens and context window.
RadOncNotes 