Hype the Hype of ChatGPT 🚀

Enough said and debated about how amazing ChatGPT is and how useful it can be across the web in the past two weeks. Since Open.ai invited the world to put on a PM hat for this new tool, I can’t resist the temptation to give it a try either.

Source: Inner-workings of the model

I’m really impressed by those ingenious experiments such as the “ building a virtual machine inside of ChatGPT “ by Jonas Degrave. I started with my immediate problems at-hand.

My six-year old asked me a few days ago: Why am I, ME?

A shot at parenting

Similar to a lot of my mommy-friends, I hunger for good advices on parenting and try my best to follow them.

But I just can’t prepare for all the edge case questions. This existential question was thrown at me several times, which means my stabs at it weren’t satisfactory. Hence, ChatGPT to my rescue.

The first time I typed the exactly question, the answer displayed as follows (took about 1 second to generate):

Okay.

How should I explain to my six year old — “why am I me?”

The last question can be seen as independent from the first one. Let’s follow-up:

The chatbot obviously remembered the context.

Regardless of whether these suggestions would work or not, they sound incredibly coherent and logical!

The end of search engines?

For my simple use case, I could also search on Google. Here’s the comparison:

I’d have to click on each of the links that interest me (Reddits, Quora and so on) to find candidate answers or look at similar questions answered in various ways.

Are search engines less effective? Maybe, if your goal is to get a coherent and reasonable answer quickly.

However, I get to choose to view different opinions, and assess different interpretations of the questions. Yes, it’s more effort to sift through the links but this is the point of doing research, it’s not just the end output that matters.

It’s the journey.

The tool enhanced my search experience, but I don’t trust it enough to completely dump Google out of my toolbox.

The end of school essays?

Saw post that says:

…the first and most common reaction I saw was fear that it would upend education . “You can no longer give take-home exams,” Kevin Bryan, a University of Toronto professor, posted on Twitter. “I think chat.openai.com may actually spell the end of writing assignments”…

I don’t believe so.

Yes it can generate something quickly. Here’s what I got prompting for today’s blog post:

Fascinating! But I can’t use it. It’s too formulaic, perhaps my prompt is not sophisticated enough — I’m not skilled at prompting yet.

But these AI generated writing gave me ideas. Here’s what I think could happen:

• Prompting becomes a skill to improve (just kidding) — But if you go back and see Jonas Degrave’s prompts, you might recognize that only domain experts can do experiments with AI like that which includes judging and assessing the quality of the generated outputs. Therefore, the more domain knowledge you have, the more sophisticated of a prompter you’d become, and hence the better results you’d get. In essence, the generative AI becomes an aid, rather than a replacement of our brains.
• The submission format of school assignments may change. For instance, I would ask my students to not only submit their essays for review, but also present it. If they demonstrate that they fully understand what they’ve written, and stand up to the challenge of tough questions (perhaps even those from an AI question generator) in front of a live audience, it’s a pass in my opinion. At the end of day, essays are just tools that help you make your case, you still have to execute effective communication in order to achieve your goal.

“Dot the eyes” of the dragon

Saw some other comments:

This technology promises to make a lot of people lazier, mostly because it’s so good at swiftly and adeptly regurgitating the rote process of wording and sentence structure that takes a normal person a laborious amount of time.

There’s a Chinese idiom called “畫龍點睛”, which is a story about a skilled painter who drew a dragon but wouldn’t finish drawing the eyes because of fear that the dragon would “fly away”. He was challenged by someone to “dot the eyes” and so he did. The dragon from the painting came to life and flew away.

I think this is where new AI tools like the ChatGPT will add value.

It will enable creators like us to “dot the eyes” and produce something even better.

At the end of the day, it’s the human beings — us who will make sense of the AI and turn them into something new and creative.

With the next-gen AI tools like ChatGPT, we will be the ones who will “dot the eyes” on the AI blueprints, and make them fly.

Happy practicing!

Thanks for reading my newsletter. You can follow me on Linkedin or Twitter @Angelina_Magr !

Quotes/Images: ChatGPT : Optimizing Language Models for Dialogue by Open AI
Blog . Building A Virtual Machine inside ChatGPT by JONAS DEGRAVE News . ChatGPT Is Dumber Than You Think by The Atlantic
Blog . How to Talk to ChatGPT, the Uncanny New AI-Fueled Chatbot That Makes a Lot of Stuff Up
News . Rise of the bots: ‘Scary’ AI ChatGPT could eliminate Google within 2 years