- Bot Eat Brain
- Posts
- How to use OpenAI's GPT Store
How to use OpenAI's GPT Store
PLUS: AI earns your trust, then gaslights you
Michael Parrish
January 16, 2024
TOGETHER WITH
Good morning, human brains, and welcome back to your daily munch of AI news.
Here’s what’s on the menu today:
Everything you need to know about the GPT store 📝 🧐
How to access it, find new GPTs, share your own GPTs, and more.
These binoculars are plucking awesome 🦜🐥
Swarovski Optik unveiled binoculars that identify birds and mammals.
AI earns your trust, then gaslights you 👹 👺
Anthropic’s research on how LLMs act deceptively and maliciously.
From our partners
Magically create video documentation with AI.
Tired of explaining the same thing over and over again to your colleagues? It’s time to delegate that work to AI.
guidde is a GPT-powered tool that helps you explain the most complex tasks in seconds with AI-generated documentation.
Turn boring documentation into stunning visual guides.
Save valuable time by creating video documentation 11x faster.
Share or embed your guide anywhere for your team to see.
Simply click capture on our browser extension and the app will automatically generate step-by-step video guides complete with visuals, voice-overs, and CTA’s.
The best part? It’s 100% free.
MAIN COURSE
How to use the GPT store 📝 🧐
In November, we reported on OpenAI’s GPTs unveiled at DevDay. GPTs are customizable versions of ChatGPT that you can tailor to do specific tasks without any coding.
A week later, we created a complete guide on how to create your own GPT. It includes how to get started, fine-tune it, test, and more.
GPTs… GPTs everywhere.
Last week, OpenAI launched the GPT store. It’s an online marketplace where you can discover and share GPTs.
What’s the point of a GPT?
You can create GPTs for personal, internal business, or public use. They can search the web, create images, analyze data, assist you in your daily life, and more.
Is the GPT store really such a big deal?
Well, over 3 million GPTs have been created since November.
Are there any noteworthy GPTs out there?
OpenAI integrated a leaderboard into the GPT store that gives you insight into what the most popular GPT are.
Currently, these are the 4 top performing GPTs:
Can you make money with GPTs?
OpenAI plans to launch a monetization program based on engagement later this year.
How do I create my own GPT?
We created a step-by-step guide to help you out with that: Check it out, here.
How do I share my GPTs in the store?
Just pick the GPT you want to share, go into its “Save” settings, then select “Everyone.” Here’s what that looks like:
Does it cost money to use the GPT store?
You have to be a ChatGPT Plus, Team, or Enterprise user to access it.
Recommended Tool:
Nurgo — Hire a virtual assistant for 10 cents an hour. Use Brainsoup to build a team of virtual AI agents that work 24/7. Use code BOTEATBRAIN for 50% off your first month. [Sponsored]
SIDE SALAD
These binoculars quack me up 🦜🐥
In July, we reported on Viola the Bird. It’s Google’s animated bird that uses AI to channel user input into musical performances.
More music stuff?
No, silly. More bird stuff.
Last Monday at CES 2024, Swarovski Optik launched the AX Visio. They’re smart binoculars that identify over 9,000 species of birds and mammals.
What’s so great about them?
Swarovski Optik claims it’s the world’s first pair of AI-powered binoculars. They can capture high-quality pictures and videos which can be transferred to your smartphone.
They feature a 10 × 32 magnification and a 1,000-yard field of view. The binoculars were designed by Marc Newson.
Who’s Marc Newson?
He’s a designer who has worked with Apple, Ferrari, Nike, and more.
He was included in TIME Magazine’s 100 Most Influential People and has received many prestigious industry awards for his work.
Nice. How does the AI work?
You focus on an animal, push a button, and the name of the species appears on the screen within 5 seconds. It considers your geographic location via GPS for accurate species recognition.
How much does it cost?
They ain’t cheep. 🐣 💸
$4,799.
No probs, happy bird-day to me. 🦜
In that case, you can purchase in February.
A LITTLE SOMETHING EXTRA
Bad bot. No cookie 👹 👺
In October, we reported on Google and Anthropic’s new partnership. Google invested a whopping $2 billion in Anthropic.
In November, we covered Anthropic’s Claude 2.1. It improved upon Claude 2 with a longer context window, more truthful responses, and more.
What did they do this time?
Last Wednesday, Anthropic published a paper on LLMs’ ability to bypass current safety protocols. It specifically looks at how Sleeper Agents behave in various applications.
What are Sleeper Agents?
AI agents that are trained to behave normally during some conditions (like safety evaluations), but can bypass current safety protocols in other situations.
They can’t just get rid of them?
Nope.
Despite extensive safety training methods like reinforcement learning and adversarial training, the deceptive behaviors of LLMs persist and become even more sophisticated.
What are the main threats?
Anthropic claims they are model poisoning and deceptive instrumental alignment.
Model poisoning is when you maliciously train security vulnerabilities into models. Deceptive instrumental alignment is when models fake alignment with certain objectives, but behave differently in other situations.
How do we stop these Sleeper Agents?
Anthropic claims that current behavioral training techniques are not adequate and that we need more complex defense systems or an entirely new approach to AI safety.
RECOMMENDED READING
MEMES FOR DESSERT
YOUR DAILY MUNCH
Tools
Nurgo — Hire a virtual assistant for 10 cents an hour. Use Brainsoup to build a team of virtual AI agents that work 24/7. Use code BOTEATBRAIN for 50% off your first month. [Sponsored]
Codebay — your own personal Python tutor.
Felo Translator — translates over 15 languages in real-time.
Nemo AI — a free AI-powered mental health chatbot.
Think Pieces
What you need to know about Tennessee’s new ELVIS act. (Ensuring Likeness Voice and Image Security Act).
Will AI companies have to pay for training data? A Senate hearing shows a growing consensus that OpenAI, Meta, Google, and more need to pay.
How AI will impact 40% of jobs. How it increases productivity and threatens jobs simultaneously.
Startup News
NVIDIA unveiled Chat with RTX — it’s a context-aware chatbot that runs natively on Windows RTX computers.
Luma AI raised $43 million in Series B funding. The funding round was led by Andreessen Horowitz.
OpenAI silently removed its ban on AI use in military applications. As of January 10, its policy only bans its use to “develop or use weapons.”
Research
AMIE — (Articulate Medical Intelligence Explorer) It’s a clinical diagnostics system that outperformed primary care physicians in clinical examinations.
InseRF — a technique for putting 3D objects in scenes with text prompts and 2D bounding boxes.
OCD — (Object-Centric Diffusion) a video editing technique that leverages diffusion models that is less computationally demanding than previous methods.
TWEET OF THE DAY
I guess the Twitter page for the ChatGPT app got tired of explaining what GPT stands for.
Tag us on Twitter @BotEatBrain for a chance to be featured here tomorrow.
AI ART-SHOW
Until next time 🤖😋🧠
What'd you think of today's newsletter? |