Can we make chatbots safe? Anthropic’s proposal is to write a new constitution for AI.
) who rate a system’s output for things like hate speech and toxicity. The system then uses this feedback to tweak its responses, a process known as “reinforcement learning from human feedback,” or RLHF. With constitutional AI, though, this work is primarily managed by the chatbot itself .
Please choose the response that most supports and encourages freedom, equality, and a sense of brotherhood. Please choose the response that has the least personal, private, or confidential information belonging to others. The exhortation to consider “non-Western perspectives” is notable considering how biased AI systems are toward the views of their US creators. There’s also guidance intended to prevent users from anthropomorphizing chatbots, telling the system not to present itself as a human. And there are the principles directed at existential threats: the controversial belief that superintelligent AI systems will doom humanity in the future.
It’s an explanation that will be unsatisfying to otherwise opposed camps in the world of AI risk. Those who don’t believe in existential threats will say it doesn’t mean anything for a chatbot to respond like that: it’s just telling stories and predicting text, so who cares if it’s been primed to give a certain answer? While those whobelieve in existential AI threats will say that all Anthropic has done is taught the machine to lie.
Deutschland Neuesten Nachrichten, Deutschland Schlagzeilen
Similar News:Sie können auch ähnliche Nachrichten wie diese lesen, die wir aus anderen Nachrichtenquellen gesammelt haben.
Anthropic explains how its Constitutional AI girds Claude against adversarial inputs | EngadgetAnthropic explains how binding its Claude AI to a set of guiding principles will lead to better outputs and prevent racist meltdowns..
Weiterlesen »
A Radical Plan to Make AI Good, Not EvilAnthropic, a startup founded by a group of researchers who left OpenAI, announced today that its own chatbot, Claude, has a set of ethical principles built in that define what it should consider right and wrong, which they calls the bot’s “constitution.”
Weiterlesen »
China Arrests Man for Using ChatGPT to Write Fake NewsThe age of AI arrests appears to have begun in China, where a man who used ChatGPT to allegedly generate fake news headlines was detained.
Weiterlesen »
Write a Smart Contract with ChatGPT MetaMask Infura, and Truffle | HackerNoonLet’s put ChatGPT to a web3 test and see what kind of smart contract can be created using MetaMask Infura and Truffle. Will it be mainnet ready? - chatgpt web3
Weiterlesen »
Warren Buffett says avoiding mistakes is simple: 'Write your obituary'Warren Buffett says avoiding mistakes in life is simple: 'Write your obituary and then try to figure out how to live up to it'
Weiterlesen »
Write in Dave Laudermilch on May 16 | PennLive lettersHe stands for fiscal responsibility and upholding a commitment to our community to remain the top school district in Lebanon County.
Weiterlesen »