I was looking to do a pen and paper brainstorming when I sat at my desk this morning in hopes I can take a small step leading to a lifelong path I want to take in terms of financial fulfillment. I know right? Tall order. And also a long sentence.

Then I remembered that yesterday out of the blue I just thought about Google Gemini and figured maybe I’ll check out what tricks it has up its sleeves because I sadly slept on it in the early days of its announcement. You can’t really blame me because let’s be honest, I’m surely not the only one who was so hyped about Google Bard when it came out and I’ve rushed to be on the wait-list expecting a mind-blowing experience only to be extremely disappointed. I may have set my expectations too high but I mean, it’s Google — it literally has access to more data than you can ever imagine so I had wild ideas of how it could use it (privacy issues aside — hello suspended reality) to better my life. Sadly though, at least for me, Google Bard was painfully restrictive and limited that I found myself crawling back to ChatGPT free version, mind you.

So anyway today I was in a let’s-give-Google-another-chance mood aka is-there-really-anything-better-at-this-point-that-I have-access-to-anyway kind of vibe. Okay, I’ll stop hyphenating now, I heard you the first time.

I’ve done these brainstorming things with LLM’s before (mainly OpenAI’s ChatGPT, but occasionally I’d go to Microsoft Bing for giggles) and let’s just say a lot of them ended in lots of hair-pulling and loud exasperation. So just like with Google Bard previously, I didn’t expect much. But I did hope it’ll be a notch better this time primarily because I’ve signed up for the free Google Gemini Advanced version. Can’t lose if it’s free so I was like let’s go Gemini! Talk to me!

I would love to ramble on and dissect the prompt exchange but I’m posting it here unedited so you have a better feel for it. I’d very much rather show than tell. And it’s also rather long because I did ask a few other questions before I dove fully into my main topic as far as it would allow for an in-depth but not too taxing brainstorming session in one sitting.

Here’s the massive screenshot of the page if you don’t like clicking links.

And oh look, Google Gemini responses also has support for audio! Too lazy or busy to read Google Gemini’s response? No problem. Click and listen!

Go on over to this link I mentioned above to read the entire chat, if you haven’t already at this point. Then come back here for some kind of article “closure” (if you’re that kind of person — not judging, I see you).

Done reading the whole thing (or part of it)? Nice. Welcome back! I hope it gave you a glimpse of what LLM’s and AI and AGI has in store for us in the not-so-distant future. It is dreadful sometimes to think that AI and most-notably AGI is bound (not out) to take our jobs and everything that’s “good” with it (uhm,hmm..) but as you can see, not everything has to be doom and gloom (as I find myself pondering on at times).

Aren’t you relieved that gone are the days of throwing spaghetti on the wall, hoping and waiting for something to stick? Or God forbid, subjecting another person in our life to such things? Sounding-off or choosing a beloved as a sounding board is so “November 29, 2022”. Surprisingly, other people have interesting ideas too (and jobs and resposibilities). And who has the time to test these ideas when everyone else is out and about living interesting lives? Not me (wink wink).

This last part is a bit jargony (is that a word?) – uhm, jargon-heavy for people who don’t care about code and so props to you if you still press on regardless.

A huge part of the criticism of LLM’s aren’t that they’re garden-variety responses or that there are a few errors here and there or that they’re not factual or that it “hallucinates” (who doesn’t eh?) — those are a given especially in the early stages. But at least in my personal experience, my issue with it is that in the beginning it can never quite “get” me. I’d have something else in mind that I’m trying to put across but the LLM seems to “misinterpret” or ignore a term or phrase altogether.

Maybe I’m the one not “prompting it well enough”.

And there lies the initial challenges of LLMs and AI in general. Because the moment you are forced to “prompt it an exact certain way” to get the “appropriate” or “commensurate” response (let alone a truly insightful one) means you have failed to design it to its purpose.

Prompts of a certain syntax are for programmers, coders and software developers. We already have that. That use-case already exists. LLMs are supposed to be for a layperson what VisualBasic was for DOS-based programmers. They gave programmers a GUI and the ability to drag-and-drop objects, hallelujah!

LLMs then are supposed to be conversational. And to use the oft-used term in LLMs/AI/AGI, more “human”. The goal is to not make coders of us all but to ultimately make the technology adapt to our humanity and not the other way around. To make it seamless, frictionless, invisible that it becomes second nature to everything we do.

We’ve come a long way from building the wheel in the hopes that one day, nothing will ever feel too heavy, too far, too cumbersome, too difficult. And in that sense, we are getting there. And who knows, maybe we’re already there. All you need to do is to notice it and bend it to your will and let it do your bidding. And in a good way, I hope — for the good of us all.