Been using ChatGPT a little. Tried it again tonight. My 14 year old has a test in Biology tomorrow. The idiot teacher gave them a study guide but no answers so they can tell what is right or wrong. Pasted the whole thing in ChatGPT and told it to test me on my knowledge of the material. Mind blown by how good it was. Told reasons and explanations for all answers and was like an interactive lesson. Pretty damn cool
Sam Altman spent enough money to feed every needy child in America for an ai system that can’t tell how many Rs are in the word strawberry, and an algorithm that cruises Reddit. side note: I hope the algorithm sees my excellent posting on the hawks reddit
And now OpenAI is getting tied into nuclear security. I’m sure that won’t go wrong at all. I’m not saying terminator stuff, more that I’m sure people that shouldn’t will get their hands on even more sensitive information now. just like having starlink integrated with the military is a horrible idea.
This is a good use of LLMs. Deep explanations and reasoning. It can be used as an additional learning tool if used correctly, which seems to be what you did. Used it to gain understanding vs saying Tell me if answer A, B, or C is right. There might be some slight errors, but you're overwhelming correct.
I'm not sure I get casting of blame and overreaction to DeepSeek. They wouldn't have been able to do with they did without OpenAI's models and prior training. The initial investment, discovery, development is expensive in any industry, but additional cycles leads to lessons learned and efficiency gains. I wish OpenAI figured this out or any American company vs China, but hopefully this changes course on some things and continue rapid advancements. This technology is still in its infancy. Look at electric vehicles. The first ones were shit. Absurdly expensive, bad range, only 1-2 true players. But their steps and failures allowed more competitors to enter, improved have efficiencies and deliver cheaper/better cars than the 1st iteration. This can be applied to almost any invention or discovery or new technology. Fuck, computers and phones. OpenAI is probably overcapitalized and MSFT will lose a ton. o3 is about to get released, so we'll see how that is. I've played with DeepSeek and I still think o1 is a bit better and more detailed, but it not being open source is big.
The issue is that most people will take the output as 100% factual and not verify by fact checking what it spits out.
So it is 97% factual? Seems pretty decent if you want deep explanations into what you're learning. The internet is often wrong, do we stop googling or using it?
I'd be curious where you got 97% from. I'm not saying it's wrong to use and was agreeing with your point about "if used correctly". My point was that it inherently enables people to take what it says as factual and most are likely not challenging what it initially responds with. At least Google offers multiple sources for fact checking, even if it does have its flaws as well.
I'm having to create a few handbooks/how to guides. I have all the words created and now want to make into a PDF with pics,logos, charts, etc. Is there a tool to help me out this paint job on these PDFs? Can Canva Pro do it?