Diving into the crazy new ChatGPT

What it is, what it does, applications for it, and changes it WILL bring.

Welp, it's official. AI is cray-cray, and it seems we're starting to hit that exponential growth curve so who knows where we'll be next year.

Funny robot dance

This week OpenAI publicly opened up the beta/testing version of ChatGPT... AND IT IS INSANE. It's open to anyone, for free, so just sign up for an account and play with it here.

You probably already know, but just in case you don't, OpenAI is the artificial intelligence company that has made huge waves in the space with the release of things like GPT-3 in 2020 (which alllll the AI writers use) andDALL繚E (the program that makes those crazy pictures based on what you tell it).

OpenAI has a lot of important people involved with it, including its CEO Sam Altman (the former president of Y Combinator). Its mission statement is "to ensure that artificial general intelligence benefits all of humanity." Originally founded as a non-profit, it changed itself to a "Capped-Profit" company in 2019 to attract investors, which seems to make sense to me. Investors can now make up to 100x their money, but no more.

The new ChatGPT is technically "GPT-3.5". It's the next iteration of GPT-3 that everyone has been using, but now it's optimized for dialogue and seems hella smarter.

So how smart is this thing? Let me show you a boatload of ChatGPT use cases that will blow your mind.

It can write poetry:

ChatGPT Poetry

It can do your copywriting

ChatGPT copywriting example

RIP recipe sites

ChatGPT Recipe Generation

Replace your lawyer

(click to watch it create a contract in 30 seconds)

Find bugs in your code

It can write code for new programs that works

Plan your PR campaign

It can help you hack a smart contract

ChatGPT exploit a smart contract

And all of this has been discovered just in the last three days since the release. Who knows what all we will find by next week?

It's probably a really good time to use this gif:

ChatGPT is not perfect though.

There is a seemingly massive amount of failed prompts, where the bot can produce wrong information, or possibly just made up information. Just scroll this search query to see tons of examples.

Also since there is not one single source of truth, there are a lot of questions that might get bad responses. The model's dataset also ended in September 2021, so it can't answer real-time queries.

It's also probably a good time to use this quote:

"With great power comes great responsibility" -Uncle Ben to Peter Parker

ChatGPT itself even warns of some of the potential bad shit that could happen.

Sinister ChatGPT Response

Considering that ChatGPT is already a better writer than me, and just a few modification scripts could send it on a misinformation campaign... you can almost guarantee that this WILL happen. Add to that some of the deepfake videos and I shudder to think what could will happen.

We already found out that ChatGPT is purposely disconnected from the internet... but it doesn't have to be. A flip of a switch and this thing is everywhere.

What can we do with this?

Again, I want to emphasize how all of these capabilities have only been discovered within the last 3 days since release. And I might add that the API access is not even live yet, all of this has just been discovered using OpenAI's dashboard prompt interface.

To take this back to GPT-3, its release was cool and all, but the real amaze-balls stuff came from allowing entrepreneurs and geeks (and entrepreneur-geeks) API access so they could build products on top of it.

So the REALLY cool stuff with ChatGPT hasn't even shown itself yet. Stay tuned for the API release.

We can guess a little though. And the ramifications could be world changing.

Here's a fun little thought for you.

For years, Google has always claimed to not really be a monopoly. They always said they were vulnerable to disruption. I always believed them on this, but I don't think too many others did.

ChatGPT could be a massive direct competitor to Google.Instead of Googling something, sorting through affiliate spam, and digging through a few websites, you could just ask ChatGPT... and have follow-up responses based on that.

Of course, we know that Google has internal projects building the same thing as OpenAI, and they have billions more in funding to play with.

In 2021, Google announced a massiveone trillion-parameter transformer system when GPT-3 was "only" using 175 billion. I am not up to date on the capabilities of Google yet (if you know please reply and share).

You can war-game out a few scenarios here just from this:

  1. Google is forced to change how its search engine business works. SEO's generally hate featured snippets, answer boxes, etc. But with ChatGPT changing how people use the internet I could see Google having to offer an "ask the AI" feature for queries that don't need an opinion. Edit* 2 minutes after publishing I found a Chrome Extension to show ChatGPT answers beside Google search results.

  2. Infinite Google-replacement competitors pop up. Below is an example screenshare of some hacker who got a version of ChatGPT working inside of Whatsapp. Need to Google something? Nah don't bother just text the AI.

(click the Tweet to see it in action)

Now multiply this over millions of script kiddies who can spin up an app over a weekend.

Makes you ponder things a bit doesn't it?

What's next?

GPT-3 vs GPT-4 meme

Let's not forget that this is "just" GPT-3.5. Just a stepping stone.

It's almost a joke at this point, but GPT-4 is next and it's expected to basically walk on water and cure cancer all while making me a margarita.

Also, try and think exponentially here. Remember humans are not naturally very good at this.

fun exponential growth vs linear growth graph

I don't have any real clue on what these types of developments will bring to our future, but I'm pretty certain they will change things.

It's also a sort of well-known phenomenon that once one person or company discovers something then others will crack the code soon after.

All of these big giant tech co's around the world with infinite budgets and infinite computing power are sure to have similar things around the corner.

Have any thoughts, ideas, or requests? Please just hit reply and say hi.