Skip to main content

Advanced Search

Projected result for your current search: 0

Search Operators:

You can use operators in your search query to narrow down your results even more. Click on the search operator to see an explanation of how it works.
Finds documents with exactly this word group, in exactly this word order and spelling (e.g. "employer branding").
Finds documents with both search terms (e.g. sales AND bonus).
Finds documents with one or the other search term (e.g. porsche OR volkswagen).
Finds documents with all search terms. Blank space is understood as AND (e.g. man robot production).
Finds documents with no appearance of the word behind NOT (e.g. ford NOT "harrison ford").
Finds documents where the search term is mentioned at least "n" times."n" may be any number (e.g. COUNT(gear)>8).
Finds documents with both search terms in any word order, permitting "n" words as a maximum distance between them. Best choose between 15 and 30 (e.g. NEAR(recruit, professionals, 20)).
Finds documents with the search term in word versions or composites. The asterisk * marks whether you wish them BEFORE, BEHIND, or BEFORE and BEHIND the search term (e.g. lightweight*, *lightweight, *lightweight*).
Finds documents with the search term in different spellings. "?" allows only one character (e.g. organi?ation).
Special characters are understood as AND (e.g. Miller Bros. & Sons).

Semantic Search

powered by (Link opens in a new window)
loading …
Log in
Top
AI & SOCIETY

Language agents reduce the risk of existential catastrophe

  • 19-08-2023
  • Open Forum
Published in:
Authors
Simon Goldstein
Cameron Domenico Kirk-Giannini
Published in
AI & SOCIETY | Issue 2/2025

Activate our intelligent search to find suitable subject content or patents.

search-config
loading …

Abstract

The article delves into the pressing issue of existential catastrophe due to the advent of Artificial General Intelligence (AGI). It focuses on the potential of language agents to mitigate this risk by addressing critical alignment challenges such as reward misspecification, goal misgeneralization, and uninterpretability. The authors argue that language agents, which function autonomously by pursuing goals specified in natural language, can substantially reduce the probability of misalignment catastrophe. The text provides a comprehensive overview of the architecture and capabilities of language agents, highlighting their ability to create complex plans and distinguish between means and ends. Additionally, it discusses the interpretability of language agents and their potential to be safer than traditional reinforcement learning systems. The authors conclude by emphasizing the importance of investing in language agent research to reduce the risk of a misalignment catastrophe.
AI Generated

This summary of the content was generated with the help of AI.

Already a customer? Then log in here to gain access.

Customer login

Not a customer yet? Then find out more about our access models now:

Individual Access

Start your personal individual access now. Get instant access to more than 164,000 books and 540 journals – including PDF downloads and new releases.

Starting from 54,00 € per month!    

Get access

Access for Businesses

Utilise Springer Professional in your company and provide your employees with sound specialist knowledge. Request information about corporate access now.

Find out how Springer Professional can uplift your work!

Contact us now

For explorers: Get to know our platforms better with a free registration.

Register for free
previous article Beyond federated data: a data commoning proposition for the EU’s citizen-centric digital strategy
next article Developing safer AI–concepts from economics to the rescue
Title
Language agents reduce the risk of existential catastrophe
Authors
Simon Goldstein
Cameron Domenico Kirk-Giannini
Publication date
19-08-2023
Publisher
Springer London
Published in
AI & SOCIETY / Issue 2/2025
Print ISSN: 0951-5666
Electronic ISSN: 1435-5655
DOI
https://doi.org/10.1007/s00146-023-01748-4
This content is only visible if you are logged in and have the appropriate permissions.
This content is only visible if you are logged in and have the appropriate permissions.

Other articles of this Issue 2/2025

The work of art in the age of artificial intelligibility

  • Open Access
  • Main Paper

Bringing older people’s perspectives on consumer socially assistive robots into debates about the future of privacy protection and AI governance

  • Open Forum

Poisoning an already poisoned well

  • Curmudgeon Corner

Automated decision-making and the problem of evil

  • Open Forum

From the essence of humanity to the essence of intelligence, and AI in the future society

  • Open Forum

Assessing the impact of heat vulnerability on urban public spaces using a fuzzy-based unified computational technique

  • Open Forum

Premium Partner

    Image Credits
    Neuer Inhalt/© ITandMEDIA, Nagarro GmbH/© Nagarro GmbH, AvePoint Deutschland GmbH/© AvePoint Deutschland GmbH, AFB Gemeinnützige GmbH/© AFB Gemeinnützige GmbH, USU GmbH/© USU GmbH, Ferrari electronic AG/© Ferrari electronic AG