Introducing Bard and ChatGPT (for novices)

Please note: I will be using artificial intelligence (AI) to help write this post. I will tell you when and where I do this. I will always use italics to show the AI response to any questions I pose.

This post is going to discuss two popular AIs that are free (so far) to the public: Bard from Google, and ChatGPT from OpenAI.

I’ll be honest and tell you that I still get a bit confused trying to figure out when something is an app vs an AI. I’ll try not to confuse you as well, but I may occasionally fail to do so. In the case of Bard and ChatGPT, these are essentially chatbot apps that use AI. Although they function similarly, they are different. BTW, one reason I get confused is I was asking Bard whether it was an app that used AI or just AI and got conflicting answers. It’s good to remember that these apps are not perfect. They can and do make mistakes (warnings abound), and occasionally they “hallucinate”. Huh? That’s right, they are known to just make shit up occasionally. Hmm. I guess that just makes them more human. LOL!

Large Language Model (LLM)

To understand what Bard and ChatGPT are, you will have to understand the meaning of the term, “large language model (LLM).” Don’t worry, a high-level definition will suffice, so I won’t bury you in tech-speak.

To save time, I asked Bard what a large language model is. Here is part of Bard’s response (in italics):

Q. What is a large language model?

A. (Bard)

A large language model (LLM) is a type of artificial intelligence (AI) that can process and generate human language. They are trained on massive datasets of text and code, and are able to communicate and generate human-like text in response to a wide range of prompts and questions. LLMs are still under development, but they have the potential to revolutionize the way we interact with computers.

Here are some of the things that LLMs can do:

Generate text: LLMs can generate text, translating languages, and writing different kinds of creative content, like poems, code, scripts, musical pieces, email, letters, etc.

Answer questions: They can answer your questions in an informative way, even if they are open ended, challenging, or strange.

Translate languages: They can translate languages, even if they are not very common.

Write different kinds of creative content: They can write different kinds of creative content, like poems, code, scripts, musical pieces, email, letters, etc.

Generate different creative text formats: LLMs can generate different creative text formats, like poems, code, scripts, musical pieces, email, letters, etc.

Follow your instructions: LLMs can follow your instructions and complete your requests thoughtfully.

Use their knowledge to answer your questions in a comprehensive and informative way, even if they are open ended, challenging, or strange.

LLMs are still under development, but they have the potential to revolutionize the way we interact with computers. They could be used to create new types of applications, such as virtual assistants, chatbots, and educational tools. They could also be used to improve the accuracy and reliability of existing applications, such as search engines and translation tools.

I think Bard’s response above demonstrates a few reasons why there is so much talk these days about AI. Translate languages? Write scripts? (There have to be some nervous Hollywood script writers out there). Write code? (Some nervous programmers, too). It’s an interesting list.

So Bard basically is a chatbot app that uses an AI model called a large language model, or LLM. And this is where it kind of blows my mind. Because an LLM is not just any model, it is a “mathematical model.” That’s right. Math. No little man behind the curtain, no budding consciousness, just numbers.

When I first read this, I thought, “Aha! So an LLM is simply a very fast predictive text machine!” meaning that it simply uses the context of a user’s input to predict the next word or phrase, like our phones do. But, I was wrong. And here I will not dive deep, because I don’t yet understand how this happens, but one of the reasons why an LLM is not just a predictive text machine, is that an LLM can be trained. Yep! It learns! (I’m still learning how it does this). The AI developers are somehow able to input a large amount of data into Bard. LLMs then use learning techniques like “beam search” to create text that is similar to what it received as input, and “attention” to focus on specific parts of the input to help it determine and create relevant responses. There’s much more, of course, but to me, so far, it’s all magic.

Note that the diet Bard and other LLMs are fed is referred to as a “data set”.  This steady diet of data (it is ongoing) and this is what is meant by “training” the AI. This training is what allows an LLM to do its thing e.g., answer questions, generate text and images, translate languages, write code, etc. The data sets that are input to Bard include books, texts, articles, code repositories, websites, essentially the entire internet. Because it’s Google, it also has access to search engine queries and responses, which allows it to give its responses a familiar look and feel.

ChatGPT, by the way, is also a chatbot app and works the same way as Bard, in that it uses an LLM, too. While ChatGPT is trained on some of the same data sets, it also consumes different sets. Bard is trained on a massive data set of text and code, while ChatGPT is trained on a dataset specifically designed for conversational applications. You’d think (at least, I did) that this would mean ChatGPT is much more friendly and human-like than Bard, but I can’t tell the difference.

BTW, this is the power of these LLms. You can have LLMs that are focused on different data sets. For instance, a travel agency could create an LLM focused on just travel, a health agency could focus on health, etc. Each would then have chatbots that worked with their own LLMs to provide services. That’s cool.

If you’re wondering how much data Bard and ChatGPT have consumed so far, well, it’s a lot. ChatGPT has been trained on a dataset of 175 billion words. Yikes! That’s a lot! Bard has been trained on a dataset of 1.56 trillion words. That’s right, trillion! All I can say is, “Holy shit!” And it’s training will never end.

Be aware also that there are two versions of ChatGPT. One is free, the other, called ChatGPT Plus, is not. I won’t go into the differences here. Just know that the free version of ChatGPT was cut off from training at the end of 2021, so its dataset is dated. This is one reason I stopped using ChatGPT. Bard, however, uses current data and is continually being trained, as is the plus version of ChatGPT. If you want the plus version, you have to pay $20 per month. There is no Bard plus version and so far, it’s free to use. I hope it stays that way.

If memory serves me right, the free version of ChatGPT also has a limit on the number of queries you can make per day.

Basic AI Terms You Should Know

Here are a few additional terms you will probably hear people mention when they talk or write about AI these days. While researching this, I decided to who better to ask, but an AI! So I queried my favorite AI app so far, Bard, from Google. Bard’s response is in italics.

Q. What are some of the basic terms we need to know to begin to understand the new AI apps?

A. (Bard):

Here are some basic terms you need to know to begin to understand the new AI apps:

• Algorithm: A set of instructions that a computer follows to solve a problem.
• Artificial intelligence (AI): The ability of a computer or machine to learn and perform tasks that are typically associated with human intelligence, such as reasoning, planning, learning, and problem-solving.
• Chatbot: A computer program that simulates conversation with human users.
• Deep learning: A type of machine learning that uses artificial neural networks to learn from data.
• Machine learning: A type of artificial intelligence that allows computers to learn without being explicitly programmed.
• Natural language processing (NLP): A field of computer science that deals with the interaction between computers and human (natural) languages.
• Neural network: A type of artificial intelligence that is inspired by the way the human brain works.
• Pattern recognition: A type of machine learning that allows computers to identify patterns in data.
• Supervised learning: A type of machine learning where the computer is given labeled data to learn from.
• Unsupervised learning: A type of machine learning where the computer is given unlabeled data to learn from.

These are just a few of the many terms that are used in the field of artificial intelligence. As AI continues to evolve, it is likely that new terms will be developed. However, the terms listed above will give you a good foundation for understanding the basics of AI.

Here are some examples of how these terms are used in new AI apps:

• Algorithms: Algorithms are used in AI apps to perform tasks such as sorting data, searching for information, and making predictions.
• Chatbots: Chatbots are used in AI apps to provide customer service, answer questions, and complete tasks.
• Deep learning: Deep learning is used in AI apps to power features such as facial recognition, image classification, and natural language understanding.
• Machine learning: Machine learning is used in AI apps to power features such as spam filtering, fraud detection, and personalized recommendations.
• Natural language processing (NLP): NLP is used in AI apps to power features such as speech recognition, machine translation, and text summarization.
• Neural networks: Neural networks are used in AI apps to power features such as image classification, natural language processing, and speech recognition.
• Pattern recognition: Pattern recognition is used in AI apps to power features such as fraud detection, spam filtering, and personalized recommendations.
• Supervised learning: Supervised learning is used in AI apps to train models that can classify data, make predictions, and perform other tasks.
• Unsupervised learning: Unsupervised learning is used in AI apps to train models that can cluster data, find patterns, and perform other tasks.

I hope this helps! Let me know if you have any other questions.

How cool is this? Bard’s ability to emulate natural language is stunning.

The response above took about 1-2 seconds and demonstrates one of the differences between how Bard is different from a Google search. Well, rather than me tell you, here is a snapshot of the response I got from Google when I asked the same question:

Q. What are some of the basic terms we need to know to begin to understand the new AI apps?

A. (Google):

Figure 1.

In Figure 1 above, I’m not sure what Google meant by stating, “With that said…”, because what you see above is a snapshot of the entire Google response. But the difference between Bard and Google seems very clear, doesn’t it?

Note Figure 2 below. Bard also tacks links onto the end of its responses in case you want to go directly to the sources where it got its information.

Figure 2

This is a very nice feature! They designed it this way because these AI don’t always get things right and if the response is important enough, it’s great to be able to check the sources.

You also see the thumb signs after each response. These are for the developers. Bard is still a work in progress. They are anxious to know when Bard gets it right (thumbs up) and when it doesn’t (thumbs down).

Bard and ChatGPT are both accessed via the web. ChatGPT has a couple 3^rd-party apps you can get on your phone, I believe. I created a shortcut on my phone for Bard and use it all the time now. It is quickly replacing Google search for me. You can also talk to Bard, as there is a microphone icon next to the input window.

Go here to access bard.

Go here to access ChatGPT.

Play with them! Enjoy!

Please follow and like me: