ChatGPT Vs Microsoft Copilot Vs Google Gemini: What Are The Differences?

ChatGPT, Microsoft Copilot, and Google Gemini are all part of a wave of generative AI models that have arrived over the past year that can take your text prompts and provide unique outputs. They can help you generate emails, process information, or research online, and have massive benefits even in accessibility contexts. With all of these systems being similar, it’s hard to pick which one is the best. So, let’s analyze the differences between the big three and which is better based on the situation.

What you need to know about AI

They don’t hate you

It’s important to note first that none of these AIs can think. They’re merely algorithms trained to chain words and sentences together based on what they’ve learned. No actions are being undertaken by these generative models. They’re just generating an output based on the processed input.

Similarly, when an AI goes off the rails and starts insulting the user, it’s not because it genuinely harbors resentful feelings. The internet is a vast place, and even thinking back to Tay from Microsoft years ago, training an AI on user-generated content online will eventually lead them to find negative corners of the internet and learn from them. As a result, bots can be mean because humans can also be mean. So Bing Chat (before it was called Copilot), wasn’t the second coming of Skynet back when it frequently insulted users.

What are GPTs?

Breaking down the confusion

Before discussing all the various chatbots, it’s important to recognize what Generative Pre-trained Transformers (GPTs) are. These are a family of language models developed by OpenAI, which are trained on huge amounts of text data and are fine-tuned for natural language processing (NLP) tasks.

The most recently available version of GPT, called GPT-4o, is an upgraded version of the GPT-4 that powers Microsoft Copilot and can be accessed by all users. OpenAI is still somewhat cagey when it comes to sharing technical details about GPT-4 and GPT-4o, but GPT-3.5, which kickstarted the AI text generation craze and powers ChatGPT, was trained on 570GB of plaintext collected from English Wikipedia, books, and web crawling.

As for Gemini, it uses Google’s proprietary model called, well, Gemini. It’s split into four: Gemini Nano, Gemini Flash, Gemini Pro, and Gemini Ultra. Regular Gemini uses the Gemini Pro model, and Gemini Advanced uses the Gemini Ultra model. There’s now Gemini 1.5 Pro as well, with an expected Gemini 1.5 Ultra to come in the future.

ChatGPT

The one we’ve all heard of

The big one that everyone knows is ChatGPT, which is arguably what got the ball rolling on AI text generation. The most recent iteration is based on GPT-4o, which is widely accessible and free for everybody on OpenAI’s website, though sometimes you may need to wait for servers to free up to generate responses if you’re on the free plan. Sometimes you’ll be kicked back to GPT-3.5 as well. ChatGPT Plus is a paid tier that guarantees you access to GPT-4o along with GPT-4, and you can also access GPT-4 through Microsoft Copilot.

The biggest downside of ChatGPT was its lack of access to the wider internet, which is why Google Bard and Microsoft Copilot were more sensible choices for someone looking to do research. However, OpenAI announced plugins to help ChatGPT interface with the wider internet, and later killed off plugins but rolled that capability into the default version of ChatGPT that you get with Plus. Copilot’s network access enables it to research with Bing by default and access URLs.

From my usage, the free version of GPT-3.5 is still pretty good, but I think it’s been well and truly overtaken at this stage. However, unless you’re using GPT-4o, it’s not the best option anymore for a free LLM. It still has an advantage in that it’s easy to use, open to everybody, great for text generation, and has an inherent first-mover advantage. However, not the best for research, but it’s a solid foundation to use when you just want to get something done quickly.

Microsoft Copilot

The one that’s everywhere

Microsoft Copilot is one of the free ways to access GPT-4 Turbo right now, and anyone can now use it. It can access the rest of the internet and will source what it finds so that you can verify the authenticity of the information it’s providing. Sure, not everyone will check the citations, but it’s better than ChatGPT, which spits information at you with no way to see where it came from. You can also choose a conversation style, asking it to be more creative, more balanced, or more precise. Even better is that Microsoft has image generation, meaning you don’t have to pay for Midjourney or set up a local instance of Stable Diffusion. Right now, this is one of the best ways to access the latest DALL-E 3 image generation model for free.

Microsoft Copilot isn’t perfect, though, and there are many limitations you need to be aware of. Because of several debacles where its predecessor, Bing Chat, would insult its users or turn a bit… weird, Microsoft has imposed daily chat limits. There are limits on the total number of messages in a conversation, which limits how much Microsoft Copilot can go off the rails. The limit is also presumably for some cost-saving reasons on Microsoft’s end.

Related

Microsoft Copilot Pro: What it is, how to use it, should you get it

If you’re looking to purchase Copilot Pro but aren’t sure how to use it, we’ve got all of the details that you need to know!

Microsoft has expanded Copilot to work on other browsers like Chrome and Safari, and it has released a dedicated mobile app that works quite well. With the recently-introduced Copilot Pro too, you’ll be able to integrate it into your Microsoft 365 subscription to generatively create slides, documents, and more. If you’re creative, then it may be worth a try.

Microsoft Copilot

You don’t need a web browser to access Microsoft Copilot anymore. The generative AI tool is now available as a standalone app on Android and iOS.

The strengths of Microsoft Copilot are its slightly more advanced language model, and you can also choose a conversation “style” to be more precise, more creative, or a balance of the two. However, it falls a bit behind because of its limitations. Microsoft Copilot is, most of all, an extension to the Bing search engine, so it’s best used for research. The “Copilot” branding is also very likely to become more and more common, given that Microsoft refers to it on both Windows and GitHub as well as an AI assistant.

Google Gemini

The one catching up

Google Gemini has seen a massive glow-up since its launch. It’s no longer limited to just a handful of regions and can be accessed throughout most of the world, including the EU. Not only has it improved its availability but it’s also just gotten better in general.

The difference between Google Gemini and its competitors is that Google is using entirely different technology to power its chatbot, as the initial technology used is based on the company’s Transformer that it first unveiled in 2017. Google’s in-house LaMDA was built on Transformer, and Bard became an extension of LaMDA that the company teased a couple of years ago at Google I/O. From what some users saw at the start, it seemed to be a bit behind what the others are doing currently. The company experienced some growing pains when it came to misinformation and fact-checking, but the situation has massively improved.

Google was surprisingly on the back foot, but it didn’t want to stay that way. Bard was advertised as a “complement” to Search but was then announced to be integrated into Google Assistant on select smartphones. It appears as if the company didn’t feel confident in Bard at first, but that seems to have changed with the launch of the Gemini language model.

Related

Google Gemini Advanced vs ChatGPT Plus: Which is better?

Both services are great and cost the same, but which is better: Gemini Advanced or ChatGPT Plus?

Gemini is now here following significant rebranding from Google, and with it comes a more confident Google. There’s now an “AI Premium” tier of Google One which comes with Gemini Advanced and 2TB of cloud storage, and it’ll integrate Gemini into the Google suite of applications like Word, Slides, and Sheets when it launches. Gemini is just Bard in different clothing, so it’ll be quite similar to what you’ve come to expect over the last few months, but Gemini Advanced is an all-new model powered by Gemini 1.5 Pro. It’s an advanced language model that you can use today.

Which should you use?

They’re all pretty good

If you’re going to use any generative AI model, then ChatGPT is honestly back to being the best around. Once upon a time, In the past it would have been Google Gemini or Microsoft Copilot, but Microsoft Copilot’s turn limit still puts me off. Gemini doesn’t have a turn limit, and with the its rebrand to Gemini, that’s became a massive leap in its favor.

What’s surprising is that Google Bard was once too far behind to even be considered an alternative right now. However, ChatGPT with GPT-4o is an all-new beast that excels at everything, and with its fast responses, citations, and code interpreter, it’s a no-brainer when compared to the rest. There’s nothing wrong with Copilot or Gemini, and if you’re ingrained in the Microsoft ecosystem then Copilot might actually be the better option for you, but ChatGPT suits me very well.

If you’re looking to save your money, then Gemini is probably the best for a free user. GPT-4o in free versions doesn’t have a code interpreter and isn’t available all the time, whereas Gemini will always work, can accept files, and can respond quickly all of the time. You can play around with the different LLMs to see what they all have, but unless you’re a Microsoft 365 user, I think Copilot is finally behind the rest of the competition now.

For what AI is currently safest, ChatGPT probably takes the crown as well. Researching new topics with AI tools like Gemini and Microsoft Copilot is a recipe for disaster, as we’ve already seen, whereas ChatGPT is great for helping you write emails or develop ideas. Microsoft Copilot and Gemini both have the benefit of at least showing you sources, but the limited search volume of Microsoft Copilot is a big negative against it, and Gemini can still simply pull up incorrect sources. Both are nowhere near as bad as they once were, but ChatGPT is a solid, reliable helper that, if you’re using a free AI, will get the job done quickly and effectively.