Huge congratulations and thanks to everyone helping bring The Naturalist to ABC: Andrew Karlsruher, Danny Cannon, Peter Jaysen, Alan Gasmer, Veritas Entertainment, ABC, and my amazing agent Erica Spellman Silverman. Excited to see Theo Cray’s world take this next step.
I've had access to @OpenAI's o1 for several weeks. My advice on using it:
1. Don’t think of it like a traditional chat model. Frame o1 in your mind as a really smart friend you’re going to send a DM to solve a problem. She’ll answer back with a very well thought out explanation that walks you through the steps.
2. Write your prompt in notepad. Plan out what you want. Explain all the steps – provide even more detail than you’re used to.
3. Use o1-mini for tasks that don’t require as much world knowledge but following things through step-by-step.
4. o1 tends to give me the parts of the answer and the complete response while o1-mini will give me the steps.
5. In my anecdotal and highly subjective experience, when you make corrections or adjustments, use patience in explaining what you need changed. Models that reason respond well to reasoned responses.
The amount of love OpenAI employees and alumni have for Sam is incredible.
Every one of us has some story about the time he took to talk about an idea or a concern.
There's just been no other CEO like that in history.
He listens.
Early on at OpenAI, I had a disagreement with a colleague (who is now a founder of another lab) over using the word “polite” in a prompt example I wrote.
They argued “polite” was politically incorrect and wanted to swap it for “helpful.”
I pointed out that focusing only on helpfulness can make a model overly compliant—so compliant, in fact, that it can be steered into sexual content within a few turns.
After I demonstrated that risk with a simple exchange, the prompt kept “polite.”
These models are weird.
We’ve spent the last few days doing a deep dive on what went wrong with last week’s GPT-4o update in ChatGPT.
Expanding on what we missed with sycophancy and the changes we’re going to make in the future:
openai.com/index/expanding-o…
I used OpenAI’s newest code model to make simple versions of games like Wordle, VR mazes and Zelda ENTIRELY through natural language.
I told it what I wanted and did ZERO editing/coding.
Post with demos: andrewmayneblog.wordpress.co…
I feel bad for the aliens when they show up. They’ll probably be expecting us to make a big deal about it.
Instead we’ll all be making exhausted sighs and rolling our eyes going “Oh great, now this.”
You might be talking to a very narrow group of people.
I spent 4 years at OpenAI and had just about every futurism conversation you could imagine.
I've come to the complete opposite opinion about AI and labor.
Sure the plow ended 99% of all jobs in Mesopotamia in 2400 BC, but you won't believe what happened next...
reason.com/2024/04/28/in-the…
We're very much in an early 1990s Internet moment.
You either use systems like GPT-4 daily and "get it", or you're sitting on the sideline calling it a fad or a bubble that's about to peak.
I feel that lots of computer scientists are having important debates over the limitations and possibilities of LLMs and related technologies, but AI is now in the wild, and having real impacts & real use cases.
Based on what I have seen, GPT-4 alone is enough for real disruption
SpaceX just announced that the first chartered trip around the moon will be filled with artists.
As a science and technology-loving geek who doesn't know Basquiat from Bisquick, I think this is....
Awesome.
Science lights the way. Great art helps us see.
#DearMoon
Because somebody entered an incorrect address I'm getting emails asking me to approve or deny requests for time off for people at some company.
Should I be a good boss or a bad boss?
I can't handle the pressure.
We've trained an unsupervised language model that can generate coherent paragraphs and perform rudimentary reading comprehension, machine translation, question answering, and summarization — all without task-specific training: blog.openai.com/better-langu…
When I was younger I used to having a reoccurring nightmare that I still had unfinished high school classes.
Now I have nightmares that I forgot to return a rental car in 2012.
I know we live in an age of wonders with smart phones, Netflix, SpaceX, Amazon Prime, etcetera, but part of me wants to live in the era when a mail order monkey was a reality.
#Iknowmailingmonksyiscruelbutcomeonmonkeys
My wife @rushbhatia and I went to an event tonight in downtown San Francisco put on by @LongJourneyVC where they’re showing The Matrix.
The event is called "Weird Wednesdays". We had no idea how weird it could get...
Before the movie they ask the audience to have a conversation with the person next to them about fate.
My wife, Rush, explains to me her thoughts on what it means to be an NPC and it had me thinking.
The movie begins and Rush and I are drawn in.
Suddenly...in the middle of the movie when the phone rings on screen the movie stops.
A real phone on stage at the front of the theater starts to ring…
I look around to see if someone is going to answer it.
Nope.
Inspired by my wife's conversation I get up and answer the phone.
A person on the other end of the phone (@cyantist) says “There’s a red pill and a blue pill. Take the blue pill and you can sit down and enjoy the rest of the movie. Take the red pill and walk out of the theater…”
I ask if I can bring my wife…
“Yes.”
I take the red pill.
We go into the lobby and Cyan and Scott Bannister are in full Matrix attire waiting.
We have no idea what is happening.
We get into an SUV with them and drive away from the theater. I'm still holding onto my Red Vines and Rush has her coffee cup and popcorn.
The conversation is pleasant, but Rush and I don't have a clue about what happens next.
Things are a bit surreal.
I don't think they want our kidneys...
We enter Chinatown and go down the kind of street where you can buy a Mogwai if you know where to look.
We pass a real estate office with a sign that says "NPC Real Estate." A coincidence, but a very on the nose one.
We get out of the SUV and on the other side of the street is a pop-up Michelin star restaurant.
We go inside and get to have once-in-a-lifetime Chinese Mexican cuisine.
But even more amazing, we spend the next several hours having a wonderful discussion about fate and free will with Cyan and Scott.
It all felt a bit out of body. At one point I turned to Rush and explained I couldn't wait to tell her about the crazy dream I was having when I wake up tomorrow.
Thank you @cyantist and Scott. Thank you to @zmwang for arranging the event and thank you @lukeeqin for the invite that started it all.
Always take the red pill. Answer the phone. And bring your wife for backup just in case...
Never let anyone discourage you from your dream!
Unless you're the guy in the apartment above me trying to learn guitar by playing the same damn Smells Like Teen Spirit chords over and over again for the last month.
Then totally give up.
One of the most important lessons I've learned is that a single idea brought to completion is worth more than a thousand unfinished ones.
Once I stopped dropping projects in search of something shinier, amazing things began to happen.
"a raccoon wearing a hoodie working on his laptop late into the night in los angeles making a 'yes' as he realizes his latest book is an Audible best seller"
abcnews.go.com/Entertainment…@OpenAI DALL-E 2.
My new novel The Girl Beneath the Sea, a mystery thriller about a police diver in Florida, is #1 on Amazon right now and is free for Amazon Prime members!
Please retweet 😀
amzn.to/2UBVYBB
I like to point out that "irregardless" *is* a word. It's not a particular good word or one that's acceptable to use in polite company.
Use it sometime and wait for somebody to say, "That's not a word." Then ask them, "What is it?"
I started as a prompt engineer at OpenAI in 2020. With ChatGPT & GPT-4, so much has changed.
We’re now in the next phase. The term “prompt engineer” is feeling as antiquated as “typist” or “computer operator.”
We’re all prompt engineers now.
I used to teach a high school program on critical thinking. Talk show legend Johnny Carson quietly funded it (as he did hundreds of other things)
A lot of successful people like @elonmusk secretly do that kind of thing - like a billionaire I know who funds poison control centers
I was walking last night when I saw a small furry animal cross my path. Me, a lover of all creatures, walked forward to identify the little guy. He had a fuzzy head and a weird face.
Confused, I stepped closer.
Then I realized I was staring at a skunk's butt.
I ran.
The end.
I have a giant inflatable human hamster ball in my garage I purchased years ago for an escape routine.
Is it time to break it out and start using that when I go outside?
In high school we would squeeze five people into my friend Manoj’s Camaro to go to lunch. Manoj was from India, Kurt was from Germany, Ali was from Egypt, Modi was from Israel and I was born in Alabama.
This was our normal. This is what I love about America.
Another interesting metric is Veo 3 costs $3.20 per 8 second clip. Sora currently gives you 100 ten-second clips per day.
That’s $320 worth of video generation per person for free right now.
“How many words should I write a day?” is a question I get frequently.
The answer is whatever you can manage. 10 or 10,000. Just keep moving forward.
200 words a day (20 minutes) is enough to write one novel a year.
(Unrelated: How much time did you spend on Facebook today?)