Original video in Chinese.
Key Takeaway
- Google Gemini offers an education discount that lets you use Gemini Advanced, NotebookLM Plus, and 2TB of cloud storage for free.
- Gemini’s enormous context window (1 million tokens) makes it outstanding at handling long documents, such as PDF translations, far surpassing other models.
- Gemini is deeply integrated with the Google ecosystem and can seamlessly handle YouTube video summaries with timestamps, Gmail email translation and replies, and content editing and generation in Google Docs and Sheets.
- Gemini’s powerful ecosystem and model capabilities give it a significant advantage in the competition among AI applications.
Full Content
Recently, it’s a great time to get Gemini. Google has launched an education discount, which gives you 15 months of free AI products, including Gemini Advanced, NotebookLM Plus, and 2TB of cloud storage.
I saw on Twitter that a lot of people have already cashed in on this wave of $300 worth of free benefits. I heard the people selling education email accounts made a killing. There are plenty of tutorials online about the specific method, so I won’t go into it here.
So once you’ve signed up, how should you make good use of Gemini? I’d like to share two experiences here, which are also a big difference between Google and OpenAI in terms of models and products. If you have good ways to use it, feel free to tell me in the comments.
First, context length.
When most models are still stuck at 128K, Gemini has already reached 1 million, and it’s planning to expand to 2 million later. Context length, put simply, is how much content AI can handle at once. So this huge context window is very valuable both for programming and for everyday use. I’ll show you and you’ll understand.
I have a PDF document that’s dozens of pages long, and I asked ChatGPT and Gemini to translate the entire thing for me.
Let’s look at ChatGPT first. After I dropped the document in, it said the document was too large and could only be translated in batches.
Gemini, on the other hand, was very straightforward and finished the whole thing in one go, and much faster too.
You see, that’s real strength, and it’s also why I like Gemini so much. It’s like ChatGPT taking one drink and still fidgeting, not even halfway done, while Gemini has already downed a whole bottle.
So in the future, for any English PDF, you can confidently hand it over to Gemini. Once you use it, you’ll understand the value of the words “full translation.”
Second, ecosystem integration.
AI has already entered the stage where applications are what matter. At this point, having an ecosystem and not having one are completely different user experiences.
The simplest example is handling YouTube videos.
A lot of YouTube videos are very high quality, like Lex’s. But his podcast is often three hours long, and I really don’t have time to sit through it. So I paste it into Gemini and ask it to summarize the key takeaway.
At this point, you can see that Gemini will call YouTube and output the key takeaway together with the corresponding timestamps. If you’re interested in a certain part, you can click the timestamp to jump straight there, which is very convenient.
By comparison, I gave the same request to ChatGPT. It was probably using a third-party plugin, but the result was much worse. First, the granularity wasn’t good enough; second, it didn’t add timestamps for jumping around.
Using Gemini to process YouTube links is a pretty frequent need. Whether you’re studying or doing content creation, it’s useful. Beyond that, Gemini has even more integration with Google’s other products.
I previously shared in a video how Gemini Deep Research, Google Docs, and NotebookLM work together. In fact, Gemini is already spread across the whole Google ecosystem.
After you become a paid user, when you open Gmail, you’ll see a Gemini chat window on the right side. You can ask it to help translate emails or draft an English reply; just tell it the general meaning.
Do you remember the document I just had Gemini translate in full? Because I had Canvas enabled, I could export the result to Google Docs. Then, in Google Docs, I could do further editing on that document.
For example, I can ask Gemini to summarize the core points of the whole text in a more easy-to-understand way. Then I can insert them directly at the beginning of the document.
Besides working with documents, Gemini can also help us work with spreadsheets. That’s too boring, so I won’t demonstrate it. Simply put, in the past we would enter an equals sign in a cell and do some addition, subtraction, multiplication, and division. Now, with Gemini, you can enter an equals sign followed by AI plus parentheses, then put in the prompt and the cells you want to operate on, and let AI handle it for you.
You can see, this is what the foundation of a long-established internet company looks like. They have models if they want models, and ecosystem if they want ecosystem. If you dare to start a price war with them, they’d be more than happy.
OpenAI, as a new company, was very glamorous in the first half. But in the second half, in the knockout rounds, it will be under a lot of pressure.
What I just introduced are all the most commonly used things. Besides that, there’s also Gem, which can be customized by entering prompts and uploading documents. Everyone can try that on their own.
OK, that’s it for this episode. If you want to talk AI, want to become a super individual, and want to find like-minded people, come join our newtype community. See you in the next episode!