December 6, 2022

ChatGPT - is only Google in danger? Or the whole mankind also?

The launch of ChatGPT, 5 ideas you can build out of it, threat to media & information

One, two, three and to the fo', Snoop Dobbyy Dobb and Dr V is at the do'...
Ain't nuttin' but a V thang, baby

This is Vixus, the weekly brief covering cool things happening in mediatech, from generative Ai to VR and beyond.

But this week's edition is solely about one thing... Can you guess what?

Snoop Dobby Dobb greets you in this newsletterizzle sizzle with some ChatGPT-dizzle.
(Snoop Dogg X Dobby from Harry Potter)
‍

Hold on to your socks, here's what's coming to you.

The launch of ChatGPT: is this the beginning of the end?
5 ideas you could build on top of this.
The massive threat to media & information.
‍

1/ A blitzkrieg went over the AI space.

On November 30, OpenAi took over the technerdland with the launch of ChatGPT.

If you were in a mid-desert meditation camp with Jared Leto and haven't been to Twitter, here's the TLDR on what the hell is ChatGPT.

It's sort of like a GPT-3 writer but better.

It's a chatbot. But not a dumb one.

You can ask questions like who were the actors in the tv series A-team?

Or you can give it prompts for writing code, blog posts, or haikus.

And it does it all.
‍

Within hours, Twitter was taken over by people asking it to...

...write code for them

...write rap songs for them

...create tables for them

...find chicken rosemary recipes without dying of the agonizing pain of scrolling through thousands of ads.

Who's a good robot - pat pat.

Nothing has felt so exciting and scary at the same time.

We might lose the AI rapper, though

As of writing this piece, OpenAi had disabled some of the creative writing functions.

Meaning it can't write ads, poems, songs, etc. (All the shitty rappers just exhaled in happiness. For Now.)

But it still seems to be able to write stories.

And it still answers questions, writes informative texts, and codes like a boss.

But the prompts we used to create Billboard 10 -worthy rap bangers or Clio-winning ads don't work anymore.

This is what the bot says:

But hey, ChatGPT, like 20-year girls, we have screenshots of the whole thing.

When asked about it, it says we are the stupid ones. It has never written any songs

Ah, just like the conversation on a morning after drunk DMing your girlfriend's bestie on Instagram…

FINAL UPDATE - before sending this, it was working again. So it might've crashed, or they were just fixing something.

Or maybe, like all rap artists, it was just being a diva.

2/ Your buddy for lonely nights.

What's also crazy is that it remembers the stuff you discussed previously in the same conversation.

So you can have a nice dialogue with it.

Or you can say, make this blog post 20% shorter, remove the last sentence. And it does all of that.

For example, if you want to know about the theory of special relativity.

You can dig deeper, ask questions about what's still confusing, ask examples.

This will drastically change how we learn things...

(chatting about special relativity)

Those midnight Googling hackathons about the meaning of life or combustion engines just got a lot much more fun.

3/ But here lies a great danger for society.

Getting those answers from the bot is so easy.

It presents them in such a confident way.

So everything it says, people will take as truth.

Especially beginners or just people not familiar with the subject matter.

There are already a lot of experts calling out the misinformation it has said.

Of course, this same problem already exists today.

People rarely go further than Google's rich snippets or first page.

But still, there's more exposure to different angles.

Or, based on the publisher, if you have common sense, you can tell which way the story is tilted.

You can read about the same thing from Fox News or CNN, but the stories are like night and day.

People know that.

But with ChatGPT, the angle isn't clear.

‍

It's a black box. We don't know how it's controlled.

So the friction of Google is what makes it better for humanity.

You're hopefully exposed to more angles and have to think about it yourself a bit.

Also, Google has a credibility system checker (that marketers are trying to abuse ofc).

And it's quite open about how to rank for the first page. And it's more "open-source information"

But a straightforward answer from ChatGPT gives us what our primal brain wants - an easy answer.

This can lead to all sorts of problems, from misinformation to a decrease in human intelligence levels.

Of course, curious people can have a proper sparring partner.

Whatever question you have, you can have a "dialogue" with it about it.

But fact-checking the information it spits out becomes critical.

‍

4/ Boom- pow-pew: "Corner of ideas" [in announcer's voice]
‍

ChatGPT lawyer

We don’t expect the Ai go to court for you, but it could be an extremely helpful tool for an average Joe to get an overview of his legal rights.

He can plug in the terms & conditions sheets or contracts and ask questions about whether I can do this or that. Or “the evil enterprise Apple says I have to do this - do I?”

If there's a predefined set of rules, ChatGPT should be able to understand what is possible, and what is not.
‍

Chatbot assistant

Let’s say you plug all of your tech stack into it.

Everything from email to invoicing system to analytics tools.

You can ask questions from it, give tasks, and it does all of it. Because it’s a “closed” circle connected together.

Like pull our revenue numbers from last month, add them to spreadsheet with margins and send to John.

There’s a new startup that’s raised $65 million dollars that’s promising to do roughly the same thing.

Based on currently available info, it seems like a chat widget always with you, it does the job you would do with your keyboard and mouse.

Not sure yet whether this could be like your techstack hub.

But one thing is for sure. It’s what Alexa wishes to be.

But in a more usable form. With more niche use cases.
‍

Chatbots that actually work.

Chatbots for customer support have been around since the birth of baby Jesus. And they never work.

Everyone hates them. Even the people making them.

But this could give new birth to them, especially for simple tasks.

And the ease and quality of use could open up chatbots to niches that have never used them, restaurants, for example.

Booking a table at a restaurant is very straightforward, could be a WhatsApp chatbot.
‍

An always-available teacher

We at Vixus are currently polishing our Javascript skills on a learning platform Codecademy.

How Codecademy works, you have a lot of bite-sized tasks with written instructions, and after a certain number of tasks are completed, you see what the final code should look like.

But too often, that isn’t enough. Sometimes you’re wrong already by then.

Or you just want to know why it works as it works. Or you want to be reminded what some functions exactly do.

You want to ask the questions, I can do it this way also, which one is better, and why?

Or sometimes you just can’t find the mistake.

Essentially you need a teacher.