Listening to very smart people talk about #GPT4 I'm reminded of the joke about a checkers-playing dog.
A guy has a dog that plays checkers. "My goodness," everyone says, "that's amazing. What a brilliant dog!"
"Not really," he replies, "I beat him four games out of five."
That's GPT4. It's capacities are amazing and completely unexpected.
But it's also so limited. You shouldn't back the dog in a checkers tournament, and you shouldn't use an LLM as a medical assistant or in many other ways.
@ct_bergstrom A dog as a living being has intelligence, but #chatgpt being a object has not intelligent. Its an inanimate object without thought, emotion or any feelings. Chatgpt is also not an #ai or #agi as many people call it, its an large language model meaning it is trained on a large languge as its base and can only answer our question from that dataset it has access to.
This is interesting. Yousaf Shah demonstrating how to use ChatGPT to build JSON for a hypothetical fast food ordering system. At the London Xojo conference.
#ChatGPT trick: When you're having a long conversation it is likely that the results will mess up. When that happens simply open another chat and ask ChatGPT to fix whatever you asked it to do
TED talk from computer scientist Yejin Choi delves into where AI shines, and where it falls short.
It's an interesting presentation that highlights some limitations of AI language models and gives some food for thought on how they could improve and evolve by trying to understand human perspective, values and "common sense".
She also makes a case for the democratization of AI and developing open AI models.
#Anki is a popular #flashcard system that helps you learn new info through spaced repetition. And you can build your own deck using popular tools. In his article, Otavio Ehrenberger helps you harness the power of #ChatGPT + #Python to create helpful study cards programmatically:
How to Supercharge Your Studying with Python, Anki, and ChatGPT
Call me what you will but if you need to ask #chatgpt or #copilot what it thinks about your code then maybe you are not a programmer and need to find a more fitting job!
EDIT: the word "hallucination" is probably out of place and misleading here, as @apodoxus argues in this thread. The anthropomorphic connotations of the word may actually create more misconceptions than help eliminate.
"“Last year at a time like this, I was getting, on average, 50 to 70 assignments, including discussions which are shorter, around 150 words each, and don’t require much research,” Collins told Rest of World. “Right now, on average, I get around 30 to 40-something assignments.""
So how many degrees should Collins have by now? Somebody snap him up.
This week on my #livecoding stream with @jitterted, we discuss estimation, responding to dealine pressure, academic studies, and more. Oh, and our tests find a bug that only #nullables could expose!
This is the final episode of the season, but we'll be back in two weeks with an interesting new problem: a natural-language role-playing game. We'll be using #TypeScript + #React with a #Java + #SpringBoot back end to integrate with #ChatGPT. Should be super interesting.
That starts in two weeks—the livestream is May 8th from 1-4pm Pacific, and the recording will go up on May 12th. Calendar reminder here:
Oh no, someone on Reddit #ChatGPT did a "Genisis by Donald Trump" and someone has turned it into a #deepFake - how long until the MAGA crowd finds it and starts using it as shitty propaganda?
A cheesy thread - can you spot the spoofed original 🧀 #CheeseGPT
🎶
Swiss and cheddar, gouda too,
All these cheeses, what to do?
Feta, brie, and camembert,
Parmesan, they all deserve
To be on the list, don't you see?
Cheese is part of history.
Back on my bullshit and poor #ChatGPT is giving me the customer service 'you are right, perhaps...' when I whine that its generated names aren't concise/snappy/cool/whatever enough