🤓 This Week in AI
Plus, 'Your Data, Your Rules': How to Manage ChatGPT’s Privacy Settings
Happy Friday!
This week, learn how to protect your data privacy by preventing ChatGPT from training on your personal information and conversations.
Also, if you’ve been interested in my AI For Non-Techies workshops but couldn't join earlier this year, I'm adding just one session for August 23rd (with an optional office hours session on 8/25.) Given my recent focus on team and company training, these individual sessions have become less frequent. If you've been meaning to join, don't miss this opportunity. You can check out the details below.
But first, here’s what you need to know about AI this week (clickable links appear in orange in emails and underlined in the Substack app)::
What do people really ask AI chatbots? The Washington Post's analysis of nearly 200,000 conversations reveals the most common uses: creative writing and roleplay (21%), homework help (17.5%), general searches (16.5%), work tasks (15%) and coding (7%). But there also were a few surprising findings:
😯 13% of prompts included “please.” At least we’re polite. Guess we're raising well-mannered robots!
😯😯 Users frequently try to bypass restrictions for explicit content. More than 7% percent of conversations are about sex, including people asking for racy role-play or spicy images. Who knew that chatbots get equal action from coding geeks and flirty users?
😯😯😯 People are oversharing personal data and information, raising privacy concerns and highlighting the urgent need for AI training.
Leaked documents reveal Nvidia, the key player behind the advanced chips that is fueling the AI boom, is extensively scraping videos from platforms like YouTube, Netflix and several other sources to train its AI models, sparking significant legal and ethical concerns. According to internal emails, the company has been downloading an estimated 80 years’ worth of video content daily.
Managers assured concerned employees raising legal issues that they had top-level clearance and approval, highlighting the AI industry's aggressive data practices.
OpenAI has a tool that can detect if text was written by ChatGPT with 99.9% accuracy, but internal debates have delayed its release. Concerns include negative impacts on non-native English speakers (which I don’t fully understand and they don’t explain) and user backlash.
One survey the company conducted of loyal ChatGPT users found nearly a third would be turned off by the anti-cheating technology. The company also claims that their current approach of watermarking can still lead to false positives and there are still methods by which bad actors could (relatively easily) bypass the detection.
Meta is offering millions to Hollywood stars like Judi Dench, Awkwafina, and Keegan-Michael Key to use their voices for AI projects, aiming to unveil these at Meta Connect Conference in September.
The actors’ voices could be featured in digital assistants or chatbots across Meta’s platforms, including Facebook, Instagram, and Ray-Ban smart glasses. The complex negotiations are ongoing and have faced hurdles with actors (and their talent agencies) seeking stricter usage terms and rights.
OpenAI sees leadership shake-up. John Schulman, one of OpenAI’s co-founders, has left the company for rival AI startup Anthropic. Meanwhile, OpenAI president and co-founder Greg Brockman is taking an extended leave through the end of the year to “relax and recharge.” There must be some serious s**t happening behind the scenes.
For businesses, these shifts highlight the importance of flexibility in your AI strategy. Building AI systems around a single company can be risky. A better approach is to build an adaptable infrastructure that can pivot to using any of the top AI models based on tech advancements, updated capabilities, cost and speed - so you can remain agile in the face of uncertainty and volatility.
WPP is partnering with chip manufacturer Nvidia to use generative AI for creating 3D ads, simplifying complex creative processes and reducing costs. This technology, tested by Coca-Cola and Ford, transforms simple text prompts like “build a giant rainforest with exotic flowers and rays of sunlight” into detailed immersive 3D environments. This partnership aims to make content production faster and more customized, all while maintaining brand guidelines.
South Korea’s new “future classrooms” will use AI-enabled textbooks to assess a reader’s reading level to adapt and personalize the content accordingly.
INSIDER TIPS
How to Protect Your Data Privacy When Using ChatGPT 🛡️
If you’re a ChatGPT user, you can now protect your data privacy and prevent the model from training on your personal information and conversations.
And still keep your entire conversation history.
Here’s how:
1️⃣ Go to Settings > Data Controls
2️⃣ Turn off the toggle for “Improve the model for everyone.”
This works for both paid (Plus) and free accounts.
I also highly recommend filling out OpenAI’s privacy request form as well. Here’s how:
1️⃣ Go to: https://privacy.openai.com/policies
2️⃣ Click on “Make Privacy Request” on the top right corner
3️⃣ Select “I have a ChatGPT account” (just make sure you have set up a free account if don’t have a paid subscription
4️⃣ Click on “Do Not Train on My Content” on the next screen
5️⃣ Input your email for verification and press “Send Email”
6️⃣ Check your email for one from OpenAI and complete the verification process (this might take several minutes or longer)
7️⃣ Confirm that the request applies to future data only and add your country of origin
And there you go! With just a few steps, you’ve stopped the model from training on your conversations.
Please share this information as broadly as possible with all ChatGPT users you know to spread the word. Most users have no idea this privacy setting even exists.
🔷AI for Non-Techies Workshop🔷
Curious about AI, but not sure where to start? You’re not alone.
AI can seem overwhelming and a bit intimidating. And none of the tools come with instruction manuals.
That’s why I’m hosting a beginner-friendly interactive LIVE-ONLY workshop that focuses on all you need to know to get started with AI and get good results.
Drawing from nearly 2 years of intensive hands-on experience, I've pulled together the most essential and practical insights to give you the knowledge and confidence to dive in – without drowning in information overload. It’s everything I wish I had when I got started (and still can’t find anywhere).
🤚 Who It’s For:
Beginners: Build a strong foundation and the confidence to get started.
Early Adopters and intermediate/power users: Level up your skills to get better results.
—
What You’ll Gain:
✅ Practical Know-How & Confidence: Empowered with all the knowledge & insights you need to begin experimenting and working with AI.
✅ Discovery of Personalized Use Cases: Uncover ways to use AI for your specific needs to enhance work or personal projects, from research and writing to analyzing complex documents. (among many others).
✅ Hands-On Experience with ChatGPT: An interactive walkthrough of ChatGPT, so you’ll feel more familiar and comfortable with its layout and features.
✅ An Essential Cheat Sheet: Your takeaway resource packed with key insights, strategies, techniques and examples from our sessions to serve as your handy reference for putting your AI knowledge into practice.
Each workshop series includes:
A live 4-hour interactive sessions (no on-demand viewing)
An optional 1-hour Office Hours session dedicated to answering all of your questions.
📅 Dates: August 2024
Session 1: Friday, August 23rd | 12:00 PM- 4:00 PM PST | 3:00 PM - 7:00 PM EST
Office Hours (Optional) : Sunday, August 25th | 12:00 PM - 1:00 PM PST | 3:00 PM - 4:00 PM EST
🪙 Investment: $329
I’m limiting the number of spots to keep our sessions more interactive and personalized.
If you're interested, be sure to register soon to secure your spot. They’re available on a first-come, first-serve basis.
If you're interested in a private group session, or one for your team or company, reach out to me directly at avi@joinsavvyavi.com.
And feel free to share this invitation with friends who might be interested.
P.S. If money is tight or you're between jobs right now, I get it and want to help. I'm convinced that AI skills are essential to navigating the shifting job market and unlocking new (and previously unimaginable) possibilities. So, just reach out, and we'll work something out.
That's all for this week.
I’ll see you next Friday. Thoughts, feedback and questions are welcome and much appreciated. Shoot me a note at avi@joinsavvyavi.com.
Stay curious,
Avi