Notion AI and the incredible progress in making generative models available to the public
Image (of course) generated by DALL-E 2

Notion AI and the incredible progress in making generative models available to the public

As I was jotting down ideas in my Notion for a new blog, I headed to?tweakers.net?for inspiration when I stumbled upon the news that Notion is launching a private alpha for?Notion AI: a writing assistant seamlessy integrated in the app. After DALL-E 2 and Github Copilot I’m once again blown away by the incredible progress being made in making generative AI available to the public this year. GPT-3 was of course already available through the OpenAI Playground, but when using this for writing I ran into two issues: loading up the Playground when you need help is a hassle, and I ran out of credits.

Honestly the only thing I can think about when writing this with a bit of writers block is “Why don’t I have access yet?”. From my experience playing around with GPT-3, these models can be incredibly useful in these situations by helping you finish that sentence or paragraph so you can move onto the next thought.

There’s a demo on Notions site as I linked above, but I’ll discuss some key features here. For example, Notion AI can generate entire blog post (drafts) for you with just a simple sentence as input, like “Write a blog post about Notion AI”. There’s also a brainstorm feature: the query “Brainstorm five ideas to promote Notion AI” results in the following ideas:

Geen alternatieve tekst opgegeven voor deze afbeelding
Courtesy of the demo video on Notion's website

But that’s not all! Notion AI also works on text you’ve already written: it can summarize your text, translate it to different languages, correct your spelling and grammar or complete half-written sentences.

The most exciting thing is not just language model integration coming to my favorite note taking app, but the speed at which these all these different technologies are being made available to the public. Despite the controversies around it, GitHub Copilot's integration in VS Code has helped me code at least 2 times faster and has shown an incredible understanding of my code, going far beyond just copying snippets from open source repositories. DALL-E 2 has saved me lots of time finding perfect (unlicensed) images for blog posts and newsletter pieces and now hopefully Notion AI can help me break out of the cycle of re-writing every sentence at least three times to make it nice to read. I’m also seeing a lot of potential for an AI powered resume writer here; I think a language model might be even better at buzzword bingo than I am.

The main reason why I’m so enthusiastic about these tools is that they allow me to use all of my cognitive capacity for creativity,?by taking the boring work (e.g. wording my thoughts correctly or creating pandas dataframes) off my hands. To use a metaphor: imagine you’re designing and building a house on your own. Instead of painstakingly laying floor tiles, hammering nails and doing electrical work, you can delegate the simple, repetitive tasks to a robot assistant that’s not as smart as you are and won’t come up with new ideas, but will diligently perform the simple tasks you assign to it. The jobs of software engineer, writer or (to a lesser extent) illustrator are becoming exercises in creativity. We’re moving into a future where an increasingly large part of your job could become writing queries and reviewing the work done by AI assistants (and especially with Copilot you?do definitely?still need the ability to review thoroughly), speeding up your work significantly and allowing you to focus on the fun parts. What’s not to like?

Johan Cahyadi

AI Web3 Digital Marketing Mentor - Harmonizing Digitalization for Better Life - Founder & CEO - Praktisi (11years), Pendamping UMKM & Instruktur Level 4 BNSP - Open Collaboration

1 年

That's great blog, Menno Liefstingh! We need to use it in many of our productivity tasks.

回复
Indy van Zijp

Creatief Strateeg | O-bureau

1 年

Great read, Menno Liefstingh! Excited to see if and how generative models get adapted by non-tech people. ??

要查看或添加评论,请登录

社区洞察

其他会员也浏览了