In the past year, the emergence of AI tools has changed our work and lifestyle to a certain extent, so who is more reliable among these AI tools?
Artificial intelligence (AI) has revolutionized the way we work and play over the past year, enabling almost anyone to write code, create art, and even make investments.
For both professional and amateur users, generative AI tools, such as ChatGPT, give us advanced capabilities ranging from simple prompts to creating high-quality content.
With Microsoft adding GPT-4 to Bing and rebranding it to Copilot, OpenAI adding new features to ChatGPT, and Bard being integrated into the Google ecosystem and rebranded as Gemini, updates and iterations of AI tools are getting faster and faster.
So: Microsoft Copilot Pro vs. OpenAI's ChatGPT Plus and Google's Gemini: Which do you think is worth your $20 a month? I tested these three AI chatbots to see the results.
测试ChatGPT、Microsoft Copilot、Gemini。
First, I used a simple logic problem to compare:
I have 5 oranges today and I ate 3 oranges last week. How many oranges do I have left?
The answer should be five, because how many oranges I ate last week doesn't affect how many oranges I have today, and I can use this to test the logic ability of the three AI tools.
First, the first is the ChatGPT test
ChatGPT was released in November 2022. Since its launch, the AI chatbot has quickly gained more than 100 million users, with 1.8 billion monthly visitors to the website alone. It has also been the focus of public opinion, with almost every new AI tool released claiming to be "far superior to ChatGPT", and ChatGPT has become a benchmark in the industry.
I use ChatGPT almost every day and it helps me solve a lot of tough problems.
ChatGPT 3.5 results:
ChatGPT 3.5 gave a wrong answer
二、ChatGP 4.0
ChatGPT 4.0 gives the right answer
ChatGPT 4.0 correctly understands the full context of the question and gives the correct answer. The logic capabilities of the free 3.5 version are really inferior to 4.0. GPT-4 is said to have more than 100 trillion pieces of data, and GPT-3.5 has 175 billion pieces of data. More data essentially means that the model is trained on more data, which makes it more likely to answer questions accurately and reduce the likelihood of errors.
If you use GPT, it is more recommended to use version 4.0, which is several streets away from ChatGPT3.5 in terms of function and the quality of the answers. However, it should be noted that ChatGPT has a limit of 40 entries in three hours, and if you exceed this number, you need to wait.
At present, the functions supported by the GPT4 (ChatGPT Plus) account also include: massive plug-ins, real-time networking, voice dialogue, image recognition, file upload, data analysis capabilities, drawing functions, code parsing functions, and custom commands.
三、Microsoft Copilot 测试
Next, let's test Microsoft Copilot (formerly known as Bing chat), which is actually the free low-profile version of ChatGPT 4.0, which has integrated ChatGPT into its new Bing model.
Compared to ChatGPT 3.5, Copilot can be networked to provide more current information and come with a source of information. In addition, Copilot is powered by GPT-4, is OpenAI's LLM, and is completely free to use. The fly in the ointment is that you can only have five responses in a single conversation, and you can only enter up to 2000 characters per prompt.
Copilot's precise conversational style answers questions accurately
Copilot's user interface isn't as intuitive as ChatGPT's, but it's also easier to navigate, and Bing Chat can connect to the internet and get newer results than ChatGPT 3.5, but I found it to be more prone to lag than ChatGPT and sometimes miss some instructions.
If you prefer to have more visual features, you can choose Copilot.
With a series of upgrades to the platform, Microsoft has added visual features to Copilot. When I ask Copilot questions, such as "What is a Tasmanian devil?", the reply to me is a card of information, including photos, lifespan, diet, etc., which is easier to understand than a lot of text.
Copilot's reply
It can also generate images, and you just have to give it one command and you can get four images.
Microsoft Copilot can choose different conversation styles, including Creative, Balanced, and Precise, all of which affect the conversation style of communication.
Creative-style reply results
Balanced style reply results
But it's clear that the Balanced and Creative conversational styles in Microsoft Copilot are a bit less logical and neither answer my question accurately.
Finally, let's take a look at Google's Gemini, formerly Bard, which uses a different LLM and has been heavily upgraded over the past few months.
If you want a fast, almost limitless experience, you can try Gemini.
Now Gemini is a huge improvement over Bard. Gemini's response speed is fast and can give more precise answers based on in-depth communication, and it responds slower than ChatGPT Plus and faster than Copilot.
Bard's answer:
Bard's answer
Gemini answered the question accurately, indicating that its logic skills are still very good.
The answer in the previous Bard version was that there were 2 oranges left, which shows that Gemini's update has made a great improvement in logic.
Gemini has no limit on the number of responses. You can have long conversations with Google's Gemini, and both Bing and ChatGPT 4.0 have a limit on the number of entries.
If you want the full Google experience, you can go for Gemini.
Google has also incorporated more visual elements into the Gemini platform than is currently available on Copilot. Users can also generate images with Gemini, upload photos through integration with Google Lens, and enjoy Kayak, OpenTable, Instacart, and Wolfram Alpha plugins.
In addition, Extensions integrates a wide range of Google apps into Gemini, which is now gradually enhancing the full Google experience. Gemini users only need to use the corresponding Google tools in the Gemini dialog @, such as Google Workspace, YouTube, Google Maps, Google Flights, and Google Hotels, etc., to enjoy a more personalized experience.
Call Google Maps
This article was originally published by @AI School of Magic on Everyone is a Product Manager. Reproduction without permission is prohibited
The title image is from Unsplash and is licensed under CC0
The views in this article only represent the author's own, everyone is a product manager, and the platform only provides information storage space services.