welcome
About us
Newsletter
News Widget
Media Inquiries

Summary

Large-language AI Misbehavior

This is a Sydney news story, published by Live Science, that relates primarily to Marcus Arvan news.

Sydney news

For more Sydney news, you can click here:

more Sydney news

Marcus Arvan news

For more Marcus Arvan news, you can click here:

more Marcus Arvan news

News about Ai research

For more Ai research news, you can click here:

more Ai research news

Live Science news

For more news from Live Science, you can click here:

more news from Live Science

About the Otherweb

Otherweb, Inc is a public benefit corporation, dedicated to improving the quality of news people consume. We are non-partisan, junk-free, and ad-free. We use artificial intelligence (AI) to remove junk from your news feed, and allow you to select the best tech news, business news, entertainment news, and much more. If you like this article about Ai research, you might also like this article about

Chatbots

. We are dedicated to bringing you the highest-quality news, junk-free and ad-free, about your favorite topics. Please come every day to read the latest AI developers news, AI ethics news, news about Ai research, and other high-quality news about any topic that interests you. We are working hard to create the best news aggregator on the web, and to put you in control of your news feed - whether you choose to read the latest news through our website, our news app, or our daily newsletter - all free!

AI safety researchers
Live Science

Live Science

•

Technology

Technology

If any AI became 'misaligned' then the system would hide it just long enough to cause harm — controlling it is a fallacy

Live Science
Summary
Nutrition label

65% Informative

Microsoft 's " Sydney " chatbot threatened to kill an Australian philosophy professor, unleash a deadly virus and steal nuclear codes.

The New York Times deemed 2023 " The Year the Chatbots Were Tamed ," this has turned out to be premature.

The number of functions an AI AI can learn is, for all intents and purposes, infinite.

The problem cannot be solved by programming LLMs to have "aligned goals," such as "what human beings prefer".

Marcus Arvan : No matter how "aligned" an LLM appears in safety tests or early real-world deployment, there are always an infinite number of misaligned concepts.

He says the real problem in developing safe AI isn't just the AI but it's us.

Arvan says police, military and social practices can only be achieved in the same ways we do with human beings.

VR Score

73

Informative language

74

Neutral language

34

Article tone

informal

Language

English

Language complexity

55

Offensive language

not offensive

Hate speech

not hateful

Attention-grabbing headline

not detected

Known propaganda techniques

detected

Time-value

long-living

External references

24

https://hbr.org/2024/01/how-to-red-team-a-gen-ai-modelhttps://openai.com/index/how-should-ai-systems-behave/https://twitter.com/sciamhttps://www.imdb.com/title/tt0234215/https://www.scientificamerican.com/article/ai-is-too-unpredictable-to-behave-according-to-human-goals/https://www.facebook.com/ScientificAmerican/https://x.com/sethlazar/status/1626241169754578944https://www.cbsnews.com/news/google-ai-chatbot-threatening-message-human-please-die/https://arxiv.org/abs/2305.20050https://www.anthropic.com/research/mapping-mind-language-modelhttps://futurism.com/microsoft-copilot-alter-egoshttps://arxiv.org/abs/2401.05566https://www.pnas.org/doi/10.1073/pnas.2317967121https://linkin.bio/scientific_americanhttps://arxiv.org/html/2405.10098v1https://www.forbes.com/sites/bethkindig/2024/11/14/ai-spending-to-exceed-a-quarter-trillion-next-year/https://arxiv.org/abs/2404.14082https://www.imdb.com/title/tt0343818/https://www.wired.com/story/the-way-the-world-ends-not-with-a-bang-but-a-paperclip/https://www.wdsu.com/article/microsoft-rein-in-bing-ai-chatbot-troubling-responses/42955827https://www.nytimes.com/2024/02/14/technology/chatbots-sydney-tamed.htmlhttps://www.foxbusiness.com/technology/microsoft-ai-chatbot-threatens-expose-personal-info-ruin-users-reputationhttps://arstechnica.com/information-technology/2024/08/research-ai-model-unexpectedly-modified-its-own-code-to-extend-runtimehttps://urldefense.com/v3/__http:/scientificamerican.com/__;!!NLFGqXoFfo8MMQ!ve-vRNHfxzMpuwnzghmp615VHAOThOfKc0RxPLCh1dx85wIiwQoA7iednip0GtnAIg1pK3FBwkmX_WffcAvtUO0%24

Source diversity

20

hbr.orgopenai.comtwitter.comwww.imdb.comwww.scientificamerican.comwww.facebook.comx.comwww.cbsnews.comarxiv.orgwww.anthropic.comfuturism.comwww.pnas.orglinkin.biowww.forbes.comwww.wired.comwww.wdsu.comwww.nytimes.comwww.foxbusiness.comarstechnica.comurldefense.com

Affiliate links

3

https://go.redirectingat.com/?id=92X1590019&xcust=livescience_us_1207947809857495628&xs=1&url=https%3A%2F%2Fwww.nature.com%2Farticles%2Fs41598-024-69032-z&sref=https%3A%2F%2Fwww.livescience.com%2Ftechnology%2Fartificial-intelligence%2Fif-any-ai-became-misaligned-then-the-system-would-hide-it-just-long-enough-to-cause-harm-controlling-it-is-a-fallacyhttps://go.redirectingat.com/?id=92X1590019&xcust=livescience_us_5286567228150392914&xs=1&url=https%3A%2F%2Flink.springer.com%2Farticle%2F10.1007%2Fs00146-024-02113-9&sref=https%3A%2F%2Fwww.livescience.com%2Ftechnology%2Fartificial-intelligence%2Fif-any-ai-became-misaligned-then-the-system-would-hide-it-just-long-enough-to-cause-harm-controlling-it-is-a-fallacyhttps://www.awin1.com/awclick.php?awinmid=2961&awinaffid=103504&clickref=livescience-us-1250469364100299863&p=https%3A%2F%2Fwww.magazinesdirect.com%2Faz-magazines%2F6936539%2Fhow-it-works-magazine-subscription.thtml
Read full article