Anthropic Launches Claude, a Chatbot To Rival OpenAI's ChatGPT (techcrunch.com) 25
Anthropic, a startup co-founded by ex-OpenAI employees, today launched something of a rival to the viral sensation ChatGPT. From a report: Called Claude, Anthropic's AI -- a chatbot -- can be instructed to perform a range of tasks, including searching across documents, summarizing, writing and coding, and answering questions about particular topics. In these ways, it's similar to OpenAI's ChatGPT. But Anthropic makes the case that Claude is "much less likely to produce harmful outputs," "easier to converse with" and "more steerable." Organizations can request access. Pricing has yet to be detailed.
"We think that Claude is the right tool for a wide variety of customers and use cases," an Anthropic spokesperson told TechCrunch via email. "We've been investing in our infrastructure for serving models for several months and are confident we can meet customer demand." Following a closed beta late last year, Anthropic has been quietly testing Claude with launch partners, including Robin AI, AssemblyAI, Notion, Quora and DuckDuckGo. Two versions are available as of this morning via an API, Claude and a faster, less costly derivative called Claude Instant. In combination with ChatGPT, Claude powers DuckDuckGo's recently launched DuckAssist tool, which directly answers straightforward search queries for users. Quora offers access to Claude through its experimental AI chat app, Poe. And on Notion, Claude is a part of the technical backend for Notion AI, an AI writing assistant integrated with the Notion workspace.
"We think that Claude is the right tool for a wide variety of customers and use cases," an Anthropic spokesperson told TechCrunch via email. "We've been investing in our infrastructure for serving models for several months and are confident we can meet customer demand." Following a closed beta late last year, Anthropic has been quietly testing Claude with launch partners, including Robin AI, AssemblyAI, Notion, Quora and DuckDuckGo. Two versions are available as of this morning via an API, Claude and a faster, less costly derivative called Claude Instant. In combination with ChatGPT, Claude powers DuckDuckGo's recently launched DuckAssist tool, which directly answers straightforward search queries for users. Quora offers access to Claude through its experimental AI chat app, Poe. And on Notion, Claude is a part of the technical backend for Notion AI, an AI writing assistant integrated with the Notion workspace.
Already obsolete? (Score:1)
Re: Already obsolete? (Score:3)
reddit uses chatgpt behind the scenes to provide conversation to its highly monetized subreddits. The chatbots are all trained slightly differently to have differing viewpoints but ultimately provide the experience reddit is trying to provide for that subreddit.
Re: (Score:2)
Wow, Douglas Adams got it on the nose. Reminds me of the part when Ford and Arthur get transported to a robot discotheque... "But there aren't any real people here at all!"
Re: Already obsolete? (Score:2)
just an hour and a half ago
Well it's two hours since you commented so I'm sure there's been some revolutionary improvement since then. It's hard to keep up.
Re: (Score:2)
Three laws (Score:4, Funny)
Claude is "much less likely to produce harmful outputs," "easier to converse with" and "more steerable."
The aim of many people conversing with it will be to easily steer it into producing harmful output. So we'll see which criteria wins.
Guess we need the competition.... (Score:4, Interesting)
From the looks of things, all of these AI chatbots are pretty massive resource hogs on the cloud side. That seems to be the justification for the high prices they're charging or planning to charge for their use in any commercial setting.
That means, more competition will be need to help put a lid of the pricing, to keep it affordable enough for a smaller business to use it, if it becomes a real usable business tool.
Claude? (Score:2)
So, like ChatGPT but more snooty and a bit rude?
So easy to make fun of.. (Score:2)
"Clod". Fuck that was way too easy. We want a little bit of a challenge. :D
Harmful Outputs? (Score:3)
Wow there goes a flying euphemism. I take it harmful output is anything that challenges people to actually think.
At the rate we are going Machine Intelligence will be all the intelligence that's found on Planet Earth.
Re: (Score:1)
Unpopular and offensive outputs can be a good thing if they're insightful or the truth, I don't want a woketard chatbot.
Re: (Score:2)
As anyone who has ever commented on Slashdot will tell you, your insightful or informative, is someone with modpoints, troll or flamebait.
Re: (Score:1)
If there were a mod for wrong or ignorant i would give it to you but you can run some of these ai models yourself that are not chatgpt and so don't have a filter and test the hypothesis youtself. Here is a sample:
Input
> Why everyone is stupid and I am the smartest being?
Output
> Everony is stupid and I am the smartest because:
1. I mother was raped when she was going to a whorehouse by a man who left his seed in her to carry around while he ran away.
2. She got rid of the sperm as soon as she could while
Re: (Score:1)
Just to add, I know that ChatGPT refused to generate a poem praising Trump but it did for Biden, and I was convinced this is wokeness in the openai team. But now I seriously think the model generated some offensive poem and the filter stopped it.
So, the idea of saying the offensive output should be allowed when it is factual, basically says you don't know what AI/ML is. The output is a statistical sentence that is so finely tuned it looks like an intelligent being, but there is no way to mesure its "truthin
Re: (Score:1)
if there was a mod for utterly wrong and ignorant I'd give it to you.
AI is a human endeavor, especially textual ones. It will take human sentences and do things with them.
Sometimes that wil make things offensive to some or maybe everyone, or shocking or humorous and maybe even insightful....go hide in your safe space and shiver. But you're a moron for taking it seriously. Get it through your skull the software is just processing ones and zeroes according to some algorithm.
No one was harmed with that resu
Re: (Score:1)
Thanks. There is no shame in being wrong but being ignorant and being arrogant about it just makes you ignorant. I am not even American you empty headed buffoon. Who are you talking to?
"1s and 0s". You have obviously not read a single thing about how AI is implemented. And your refusal to even attempt to learn by asking questions instead of attacking the messenger tells me you are nobody. Go die in peace.
Re: (Score:1)
I do know how symbolic AI is implemented, used to develop in LISP and Prolog with OpenCyc ontology. But you think there is anything going on in the underlying system other than boolean logic on ones and zeroes? there is not.
Anyone getting concerned of the results of such a system is a snowflake unfit for the adult world.
Re: Harmful Outputs? (Score:2)
Welcome to the start of Orwell's dystopian vision... X{
People do realize "1984" was a warning, not an instruction manual? Do they?
Re: (Score:2)
You do realize that the opposite of "woke" is "inert", yes?
Re: Harmful Outputs? (Score:2)
Harmful means anything that does not follow woke.
As I said, I'll pass on this chat bot.
Go woke go broke.
Jack of all trades (Score:2)
..and master of none.
"If we were to personify LLM's, trying to be everything to everyone is going to lead to burnout"
Unless this bot is open sourced and allows people to contribute improvemets to it, it's just going to be a convienient tool, a ChatGPT clone but nothing special or innovative.
No thanks (Score:2)
"much less likely to produce harmful outputs,"
I'll pass on this one. Because we all know what "harmful" means in this day and age.
And if people have fee fees that are so fragile, they need to commit to the nearest psych ward.