Good morning friends! ☀️
Before we crack on, I want to announce a small adjustment to this newsletter. I have decided to remove the “What has been cooking section" as I feel most readers are interested in the main body and the news part is a bit redundant. Not to mention, there are so many great AI newsletters out there. However, please tell me if you disagree!
Anyway, this week, I want to discuss something that has been debated consistently for the past year, will AI take your job?
![](https://substackcdn.com/image/fetch/w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F81c52305-a6e5-4607-8c29-14c2262cfbed_3000x1300.png)
Since ChatGPT's release in November 2022, there have been widespread news articles, blog posts, interviews, and tweets stating that “AI will take your job in the next few years.”
Now, no one is ever certain, and I am not going to tell you one way or the other. However, I want to give you my opinion, which I think is "fairly" valid as I work directly in the industry, know how the algorithms behind the current AI systems work, and use these tools daily.
I also apologize if the following sounds like a rant.
I want to start by stating that these AI algorithms are not “new.” The model behind ChatGPT is called GPT-3 (now GPT-4). These models are Large Language Models or LLMs. The first GPT model was released in 2018, and probably no one raised an eyebrow.
Sure, five years may not seem long, but GPT (generative pre-trained transformer) wasn’t just plucked out of thin air. It came from something called a recurrent neural network invented in 1990, which itself came from the neural network that originated in the 1950s.
AI is not a recent thing, it's been around for decades. It's only due to the computing resources and the internet nowadays that we have so much data that we can feed into these models. Also, let's not take anything away from the product itself. The main advantage of ChatGPT is its usability and availability to the consumer.
However, these models are only as good as the data you give them. Garbage in equals garbage out. For example, many text-to-image generation products like DALL-E and Midjourney can only produce clock faces that are 01:50 because of the data they were trained on. ChatGPT, in the beginning, didn’t know anything that happened after 2021.
Our current AI models do not possess any human reasoning or critical problem-solving skills for the reason I mentioned above. Crudely, all they are doing is retrieving information to answer your question (not completely the case as they are generative). You can just as easily find that information on the internet, it may just take longer.
As Yann LeCun puts it:
![](https://substackcdn.com/image/fetch/w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F7e9612bd-a1ce-4fdc-8a75-e7b49143633f_1162x390.png)
Yann LeCun is VP & Chief AI Scientist at Meta and won the Turing Award in 2018, the computer science equivalent of a Nobel Prize.
The reason I mention all these things is to demonstrate that AI has been in our lives for a long time. There has never been a worldwide consensus that it will replace us.
Technology does improve over time, and I am not saying that some jobs may not become redundant. However, the transition will be slow, and we will adapt promptly that you probably won’t even notice.
This recent AI craze reminds me of the blockchain hype ten years ago. How many people do you hear talking about blockchain now?
My line manager asked me an interesting question. So, ChatGPT was released on 30th November 2022, nearly 1.5 years ago, but how much has it changed your life?
The chances are not that much. You still have a job, most likely the same one, you probably use the same tools, and maybe you use GenAI tools once a week.
This space is quite nuanced, I am sure people are going to disagree with some or all of my points. However, this is just an opinion from someone irrelevant in affecting how this technology will play out. I would love to hear anyone else’s thoughts on this!
Weekly Favourites ❤️
📚 Book — Invisible Women by Caroline Criado Perez. This is my current reading, which is about how conscientiously and unconsciously the world has been built for men and ignoring half the population. The author also runs a newsletter on substack: Invisible Women.
🎙️ Podcast — Demis Hassabis: Scaling, Superhuman AIs, AlphaZero atop LLMs, Rogue Nations Threat. A friend of mine recommended Dwarkesh Patel’s podcast. It’s very similar to Lex Fridman’s and this episode was terrific. I am a big fan of Hassabis, mainly because the AlphaGO documentary is what got me into data science!
🎬 Video — Harvard CS50’s Web Programming with Python and JavaScript. Working through this course to build my website! Recommend to anyone who wants to learn web dev.
(PS: Some links are affiliate links that I get a kickback from with no extra cost to you 😎)
My Latest Content 🎬
Also, If you haven’t done that already, you can:
Follow me on Linkedin, X (Twitter), or Instagram.
Join the 8K+ beautiful people on my Medium blog where I post technical articles.
Check out my YouTube and join 4K+ people learning about data science.
💡 If you are interested in sponsoring this newsletter see here.