Weekly AI Pulse #51, Midjourney’s New Editor, Google’s AI Updates, and Grammarly’s AI Detector!
Your Best Friend to Catch Up AI News
With this week’s AI News, we’ve been publishing AI updates for almost 1 year. During this time, many of you have used these newsletters to get ahead of your colleagues or create businesses that generate significant income.
To celebrate our 1st year on Substack, we’re offering a 40% discount if you become a paid subscriber for a year now ($4 per month). Here’s the link for you to take advantage of this offer!
If you want to know more about why you should subscribe for just $4 a month, check out one of our series below. This week’s newsletter is available to everyone, so let’s dive in!
Listen this one
Fine-Tuning Now Available for GPT-4o
OpenAI has introduced fine-tuning for GPT-4o, enabling developers to customize the model for specific applications. This feature allows developers to refine the model’s tone and structure and the ability to follow complex instructions with minimal training data.
As part of the launch, 1 million free training tokens are offered daily until September 23. Fine-tuning has significantly improved across multiple domains, including coding and text-to-SQL benchmarks. OpenAI emphasizes data privacy and safety in the fine-tuning process.
Listen
Google’s AI Overviews Get Three Key Updates
Donald Trump recently posted AI-generated images falsely depicting Taylor Swift and her fans as supporters of his presidential campaign. These images originated from the John Milton Freedom Foundation, a small Texas-based nonprofit with ties to right-wing media influencers. The group promotes itself as defending press freedom but is primarily engaged in spreading engagement bait and misinformation on platforms like X (formerly Twitter). The incident highlights growing concerns about AI's role in election disinformation and right-wing propaganda.
Listen
Grammarly Introduces AI Content Detector: Here’s How It Works
Grammarly is launching a new AI content detection tool called Grammarly Authorship. This tool aims to help users identify whether content was written by a person, generated by AI, or a mix of both. While it’s available to everyone, Grammarly is particularly targeting the education market, where AI detection is increasingly critical.
How Grammarly Authorship Works
Unlike many existing AI detectors that rely on algorithms to spot AI-generated text, Grammarly Authorship takes a different approach. The tool tracks the entire writing process in real-time. It can differentiate between text typed by a human, content copied from another source, and AI-generated sections. This ability to monitor live typing gives it an edge in accuracy.
Once active, the tool categorizes the content into several groups:
Typed by a human
Generated by AI
Modified with AI
Pasted from a known or unknown source
Authorship’s analysis includes a breakdown of these categories, with a color-coded report showing how much of the document falls under each category. There’s even a replay feature that shows how the document was written over time.
Key Features and Availability
Grammarly Authorship will launch in beta next month in Google Docs and expand to Microsoft Word and Apple Pages later this year. The tool will work across all Grammarly plans, including the free version.
One standout feature is the ability for students to use Authorship as proof of their work. In cases where AI-generated plagiarism is suspected, students can show how they wrote their paper, which could resolve disputes with educators. Additionally, early next year, the tool will prompt users to properly cite content sourced from external references.
Why This Matters
As AI tools become more prevalent in education and professional settings, detecting whether content is genuinely authored by humans has become a hot topic. Other companies like OpenAI have attempted similar tools, but many have struggled with accuracy. Grammarly hopes its real-time tracking approach will set it apart and provide a more reliable solution.
The Bottom Line
Grammarly Authorship aims to bridge the gap between AI usage and transparency, particularly in education. As more schools and businesses adopt AI tools, this feature could play a crucial role in ensuring clarity and trust in content creation.
Listen
Recommended Articles
Russia’s AI tactics for US election interference are failing, Meta says
Microsoft PowerToys will launch and automatically arrange your favorite apps
Google quietly opens Imagen 3 access to all U.S. users
Listen to Summaries of Recommended Articles
Russia’s AI tactics for US election interference are failing, Meta says
Microsoft PowerToys will launch and automatically arrange your favorite apps
Google quietly opens Imagen 3 access to all U.S. users
Final Thoughts
Thanks for reading or listening to this week’s AI News.
See you next week!