Articles

Explore

Extracting ideas with ChatGPT

I just found one role emotions have in learning

My first emotional response when hearing about ChatGPT

During the hype sparked by the release of ChatGPT, I wanted to learn more about this new tool everybody was talking about.

People on the 2 platforms I was using extensively, LinkedIn and YouTube were all posting different things that they experimented with ChatGPT. It seemed that everyone was talking about how ChatGPT will end our careers and how cool and awesome this was. I know it was mostly an attention grabber. I also saw other, more technical people post all kinds of things they managed to get the AI to spit out, from code to excerpts of complicated scientific subjects.

I didn’t share their hype and enthusiasm. I mostly felt uneasy, worried, and curious. I’m highlighting these emotions because they are central to this article and will reveal how soon.

It took me a few weeks to try ChatGPT myself. In fact, I first used it to help me write this article - not as you may think, though. I didn’t use ChatGPT to write any words in this article. Instead, I used it to generate a summary of the transcript of the video below:

⁠

Comparing my insights with ChatGPT’s insights

Below is the summary that ChatGPT generated for me. Due to the length of the transcript, I had to split the transcript in two, generate a summary for each part and then summarize the two summaries.

Summary 1

Reid Hoffman and Sam Altman discuss the business opportunities that will arise from large language models and APIs. They discuss how companies can create unique businesses by tuning existing models and creating a middle layer of value. They also discuss how AI will accelerate scientific progress, with AI-dedicated products, tools that make scientists more productive, and AI that can help solve the alignment problem.

Summary 2

In this transcript, Sam Altman and Reid Hoffman discuss the potential of AI in the next couple of years. Altman predicts that language models will go much further than people think, and that we will get true multimodal models working. He also believes that AI will become the biggest technological platform of the generation, and that it will seep into almost every aspect of life. Altman also predicts that the cost of intelligence and energy will trend rapidly towards zero, and that this will have a seismic shift on the cost structure of society. Lastly, Altman believes that the metaverse will be a new container for software and a new computer interaction thing, and that AI will be a legitimate technological revolution.

Summary of the first 2 summaries

Reid Hoffman and Sam Altman discuss the potential of AI in the coming years and the business opportunities that will arise from it.

They believe that large language models and APIs will create unique businesses by tuning existing models and creating a middle layer of value.

They also discuss how AI will accelerate scientific progress, with AI-dedicated products, tools that make scientists more productive, and AI that can help solve the alignment problem.

Altman predicts that language models will go much further than people think, and that we will get true multimodal models working.

He also believes that AI will become the biggest technological platform of the generation, and that it will seep into almost every aspect of life.

Altman also predicts that the cost of intelligence and energy will trend rapidly toward zero and that this will have a seismic shift in the cost structure of society.

Lastly, Altman believes that the metaverse will be a new container for software and a new computer interaction thing, and that AI will be a legitimate technological revolution.

Why I did this exercise

The reason I wanted to use ChatGPT to generate a summary of the transcript was two-fold:

First reason. You see, the most liked comment of the video is from a person who added the summary of the transcript of the video generated by ChatGPT. I wanted to test and see if the summary generated by ChatGPT for me was anything similar to the one generated in the most liked comment of the video.

Here are both summaries where I manually highlighted similar ideas. Spoiler — There’s just one common idea in the 2 summaries generated by ChatGPT.

ChatGPT summary posted in the video comment

Large language models will create opportunities for new businesses to challenge Google in search products and provide human-level chatbot interfaces for medical and educational services.

Multimodal models and agents using language interfaces to complete tasks will also be significant trends, and successful businesses will focus on creating something unique and differentiated by fine-tuning existing large language models for specific purposes.

Access to data will be key to success in the field of AI, and businesses will need to consider data privacy and security as well as how language models can be used to improve existing products and services, create new products and services, automate tasks, and provide personalized experiences for customers.

My aggregated ChatGPT summary again

Reid Hoffman and Sam Altman discuss the potential of AI in the coming years and the business opportunities that will arise from it.

They believe that large language models and APIs will create unique businesses by tuning existing models and creating a middle layer of value.

They also discuss how AI will accelerate scientific progress, with AI-dedicated products, tools that make scientists more productive, and AI that can help solve the alignment problem.

Altman predicts that language models will go much further than people think, and that we will get true multimodal models working.

He also believes that AI will become the biggest technological platform of the generation, and that it will seep into almost every aspect of life.

Altman also predicts that the cost of intelligence and energy will trend rapidly toward zero and that this will have a seismic shift in the cost structure of society.

Lastly, Altman believes that the metaverse will be a new container for software and a new computer interaction thing, and that AI will be a legitimate technological revolution.

Second reason. I wanted to see if the summary generated by ChatGPT reflected or included the ideas and insights I gained after watching the video. I really wanted to see if ChatGPT was a tool I could use to quickly extract ideas from lengthy videos and, extrapolate, possibly lengthy pieces of text.

My own unattended and unaided by any tool (AI or not) summary of the video:

Business Opportunities with APIs: There are many business opportunities that can be created with the use of APIs, such as copywriting, education services, and medical services.

Multimodal Models: Multimodal models will open up new possibilities for businesses, such as a human-level chatbot interface and agents that can use computers to do things for users.

Science Acceleration: AI can be used to accelerate scientific progress, such as with AlphaFold and tools that make scientists more productive.

Alignment Problem: The alignment problem is the challenge of creating AI that does what is in the best interest of humanity and avoids misuse. Self-improving systems may help to solve this problem.

Language models are expected to make significant progress in the next couple of years, with the development of multimodal models and models that can continuously learn.

AI has become a buzzword, and there is a risk that the field may fall apart. It is important to look for the next paradigm and make predictions based on what is likely to work.

AI is expected to have a major impact on many areas, including financial markets, and the cost of intelligence and energy is expected to trend toward zero.

The intersection of the metaverse and AI is uncertain, but it is likely that the metaverse will fit into the new world of AI rather than the other way around.

I should learn more about AI and how I can leverage it in my business

Sam Altman is smart and I want to learn more about him and what he does

Reid Hoffman had really good questions and I want ot learn more about him

Interestingly enough, my own summary (done before I used ChatGPT to do the summary) had more in common with the summary ChatGPT generated for me. The ideas of large language models will create many business opportunities, cost of energy and intelligence going toward 0 and the idea about metaverse and AI were common.

My findings and the big ‘aha’ moment

I was pleasantly surprised to see common ideas popping up in my and the AI-generated summaries. This suggested that there is real value in using AI to generate summaries. An AI-generated summary can align with yours. This is awesome and many have shown even better and more interesting ways to use ChatGPT.

I was even more surprised to learn a few things about myself and maybe about us, humans, in general.

Intention is a big factor in learning and forming ideas

All these realizations happened because I wanted to see the difference between what AI and I can come up with. I’m pretty sure this article would have looked very different if my intention was to see how much better ChatGPT is at coding compared to me

2. Learning is contextual and deeply personal

I watched the video with my then current interests and background. That’s why one idea I extracted is that I should use AI in my business. Being interested in building a great business, it was only natural that I thought about AI within the context of my business.

And...my biggest finding

Emotions drive learning

You probably noticed that I extracted a couple of ideas - specifically about Sam and Reid that are fundamentally human. Saying that Sam is ‘smart’ and Reid ‘asked good questions’ are exactly the type of value judgments that only we can make and we probably don’t want an AI to make - I’ve recently seen a

tweet⁠

where ChatGPT refused to write a poem of praise for Donald Trump but it agreed to write one for Joe Biden and it’s a good example why we don’t want AI to make these kinds of value judgments.

These value judgments are exactly the things that make us biased and make us idolize others, but they are also the things that drive us to learn more, make connections, and pick up models - for better or for worse.

I realized after the fact that these impressions these 2 left on me drove me to learn more about OpenAI, to learn more about Greylock, startups, YC, find other useful interviews with Sam, and more.

In short, these emotions and value judgments helped me create a kind of neuronal network that helped me learn more and not necessarily more about ChatGPT or AI, but also about other interconnected subjects, like startups.

I realized that I was able to use the emotions I mentioned at the start of the article - uneasiness, worry, and curiousness - to correlate what I heard with my previous experience, contextualize the findings, look at the metadata of the interview (the 2 interlocutors) and help me search and learn more.

I imagine that one of the unrealized potentials of AI Sam was hinting at in the interview is the fact that AI will be able to create these kinds of connections, look at the metadata and related subjects of a given piece of information and create a similar neuronal network and even an entire map of interconnected things for us to explore starting from just one topic.

Nevertheless, isn’t it great that emotions play such an important role in learning? It’s a double-edged sword like almost everything else.

P.S. I’m sure this is not new knowledge for some (psychologists maybe) and I’m just discovering another thing many others have discovered before. It’s damn fulfilling to arrive at a piece of knowledge ‘on your own’ like this, though.

This is maybe something that AI cannot do for you. I have a feeling (notice I’m talking about emotions again) that fast information - the kind generated by summarizing a subject with AI - will not be able to imprint in us the knowledge and abilities we strive for and need.

Want to print your doc?
This is not the way.

Try clicking the ⋯ next to your doc name or using a keyboard shortcut (

CtrlP

) instead.