Chat GPT

ChatGPT is like a child, and that should comfort you!

Table of Contents

Wow, did we just say that ChatGPT is like a child!? Why is something that can do such amazingly smart things and learn and improve on the go being compared to a child? Is it because the new possibilities and the disrupting aspect of it keeps us awake at night and does a lot of stuff that we don’t want it to do? It’s much easier than that. It’s just a metaphor, but a strikingly fitting one.

ChatGPT is like a child because just like with children it has to be educated. Everything it does, was learned at one point. It just needs the right input to learn and after that the right feedback. The cool answers it gives to questions you ask? It found somewhere that those answers were the correct ones to the same sort of questions you’ve just asked so it should be the right answer. Writing a blog in a certain style that you like is just copying the words it attaches to that style from another source. So ChatGPT is basically a good resemblances of processing faster what we already have established and serving it at the right moments. It’s learning everything just like a child.

AI needs education too

Like any other machine and AI related product it therefore needs proper input. The software (which runs the algorithm) needs good input to learn. Actually, it needs good data from a good teacher. That is what is at the core of every AI and machine learning related project. You need good data to have crystal clear input for the model. Every flawed dataset that you use is like teaching a child that 2×2 = 4 and 2×3 =5. If this is the first interaction with mathematics, the child will see  wrong relationship between the numbers as the truth and it’s hard to adjust. So basically ChatGPT is like a child that has had a decent amount of well-done education. However, there could be a lot more to come if we send it to high school, University and maybe even let it get a PHD.

So why should this comfort you? Well, AI is not taking over our jobs. It could make them more efficient, faster or scalable but we still need to teach it. Someone needs to see if the input is correct. Give feedback to the system based on the output and adjust and improve all the time. Just like a teacher does. How often are data professionals in a meeting where there is a problem and at some point someone throws in AI and that’s the solution to all your problems. The people that throw in this solution don’t know how it works and don’t know what it takes to get there or what fundamentals are needed. They are not aware of the full education for an AI model that is needed.

ChatGPT

New revolution

Now that you know that AI is not the core solution but good input and data is, we know that there is a whole lot of work for us to do. So we will not lose our jobs, they just might change. Just like machines didn’t take away jobs. They changed our jobs and made our output increase. We’re on the brink of a new revolution and that child has just started to grow up! Are you willing to teach it just a bit more life lessons?

Follow The Data Story on LinkedIn to stay up to date on our blogs and more!

More Data stories

Analytics for a Better World
Data stories

Analytics for a Better World

I, Sophie, attended the Analytics for a Better World (ABW) annual conference at Amsterdam Business School on May 24th. This event brought together speakers and panelists from different groups: nonprofits,...
GA4
Data stories

GA4 Data Alerts in Slack

In this blogpost you will read about a Data Alerts system in Slack: a Slackbot that alerts users when it finds inconsistencies in data, using predefined BigQuery queries. After a...
a/b test
Data stories

A/B testing: In statistics we trust

Although the simple mention of the word ‘statistics’ might make your skin crawl, statistics are an essential tool of A/B testing. We created a simple cheat sheet to help you...
time-on-site
Data stories

Don't focus on time-on-site

At The Data Story we try to make data understandable. To us, that is one of the most important aspects of turning data into gold. Once you understand what the...
Product experimenting
Data stories

Product experimenting: which test should you choose?

Navigating the world of Product experimenting can sometimes be challenging. Which test should you choose? How many users should you have in it? How do you make sure your results...
Attribution modeling explained
Data stories

Attribution Modeling explained:

Since we are a Data Agency, we get many requests about Attribution Modeling. We agree, it’s definitely possible to make a lot of fancy models using data. In many cases...
nl_NLNederlands