🧙🏼 How pro writers use ChatGPT

Howdy, wizards.

Dario’s Picks

The most important news stories in AI this week

1. OpenAI shows how professional writers are using ChatGPT. AI is a powerful writing tool – not just as a tool to write for you – but to extend your creativity.

    

   Here’s how the pros use AI to write:

   • Editorial feedback: Writing is solitary – which is why editor’s are needed to bring new perspectives and give pointers on what works and what doesn’t. Asking ChatGPT to evaluate your writing is a great way to improve it.

   • Wordfinding: A thesaurus is great. But it doesn’t have context. ChatGPT can, based on the context of your writing, find the word that articulates your idea with the greatest precision and clarity.

   • Reverse interviewing: You can prompt ChatGPT to ask you questions, drawing out insights to better articulate what you want to say. Here’s an example from Stew Fortier (honestly mind-blowing).

   • Comedy: Context makes things funny. ChatGPT can help you research joke setups, common themes and exaggerated observations. It’s not going to be funny by itself, but it can help you find see patterns and points of difference in a context that you can craft into a punchline.

   • Worldbuilding: ChatGPT can accelerate the time it takes to research new topics when writing. It lets you quickly dive into any subject – allowing you to make discoveries and get quick answers on specific questions. You can also ask for sources to verify the information. And you can ask for targeted reading material if you’d like to go deep on a subject.

• ‎ Why it matters‎ ‎ Here’s what marketers of AI writing tools don’t really tell you: good writing is a difficult task, and AI isn’t going to do it for you. But, it can help you do it a lot faster and better. You just need to use it in a way that doesn’t undermine your voice, inspires you and/or makes tedious parts of writing more effective (like research).

Continued after the ad… ‎

This issue is brought to you by

“I’ll just do it myself.” Sound familiar?

When it comes to administrative tasks, hiring someone to handle them may not be at the top of your to-do list.

But the truth is this: If you don’t have an assistant, you are the assistant.

Still, your time is too valuable for you to afford your own hourly rate to keep owning those tasks.

You need to let those tasks go. You need to get out of your way. You need to delegate tasks that someone else could do for you.

BELAY is the partner you need to help you tackle those important but time-consuming tasks on your to-do list.

BELAY’s flexible staffing solutions help you reclaim your valuable time with exceptional, highly vetted Virtual Assistants who will step in to handle those frequent, time-consuming tasks.

And in as little as one week, you can be intentionally matched with a BELAY Virtual Assistant – with a list of 25 things they could start to own for you on day one – so you can get back to what only you can do: Growing your organization.

Learn more here.

2. ‎Replit Agent available in early access. Replit is an AI-powered software development platform. Replit Agent helps you create and deploy apps based on prompts, using the tools that already exist on the platform. Things like setting up a development environment, installing packages, configuring a database and deploying it to the web are now absolutely feasible for non-technical people.

• ‎ Why it matters‎ ‎ For years, there’s been a movement of low/no-code with tools that let you do stuff you normally would do with code. So far, these have been limited by platform-specific quirks and limitations. However, with AI, you can now not only write apps using code that the AI automatically wrote for you – it will also set everything up neatly and publish it for you as well!

From our partners

200+ hours of research on AI tools, prompting techniques & hacks packed in a solid 3 hour masterclass.

4. Reflection 70B, an open-source model claiming world-class performance is accused of fraud. A New York startup named Hypewrite (previously OthersideAI), co-founded by Matt Shumer, recently released a model based on Llama 3.1 with impressive benchmarks claiming to be “the world’s top open-source model”. However, third-party evaluators weren’t able to replicate the alleged performance, fuelling accusations on X about fraud. Some even speculate it could just a wrapper built around Anthropic’s Claude 3 model.

    

• ‎ Why it matters‎ ‎ Lots of red flags in this story. Gaming LLM benchmarks is apparently not so difficult. Until a model makes the ranks on something like the LMSYS leaderboard (which is based on user votes) and is verified by an independent third-party tester, it’s probably good to be remain skeptic to these results.

5. *Love Hacker News but don’t have the time to read it every day? Try TLDR’s free daily newsletter

    

    

   *sponsored

    

Affiliate disclosure: To cover the cost of my email software and the time I spend writing this newsletter, I sometimes link to products and other newsletters. Please assume these are affiliate links. If you choose to subscribe to a newsletter or buy a product through any of my links then THANK YOU – it will make it possible for me to continue to do this.