Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • NEWS AND VIEWS

Online tools help large language models to solve problems through reasoning

The large language models popularized by chatbots are being taught to alternate reasoning with calls to external tools, such as Wikipedia, to boost their accuracy. The strategy could improve fact-finding outcomes, as well as online shopping.
  1. Aleksandra Piktus

    1. Aleksandra Piktus is at Hugging Face, New York City, New York 11201, USA.

    View author publications

    You can also search for this author in PubMed  Google Scholar

As the success of ChatGPT shows, large language models (LLMs) are becoming ever easier to use. The power of these systems lies in their ability to respond to textual prompts by generating natural-sounding language. But what makes them even more effective is their ability to learn by example: by taking in just a few demonstrations, LLMs can markedly improve their performance on complex tasks1. Precisely how to formulate these examples to elicit accurate answers is an open problem, but researchers have some ideas. Reporting at the Eleventh International Conference on Learning Representations (go.nature.com/42qwbwg), Yao et al.2 propose ReAct — a prompting strategy that improves on existing methods by breaking down a multifaceted reasoning problem into bite-sized tasks and outsourcing them to external tools.

Access options

Rent or buy this article

Prices vary by article type

from$1.95

to$39.95

Learn more

Prices may be subject to local taxes which are calculated during checkout

Nature 618, 465-466 (2023)

doi: https://doi.org/10.1038/d41586-023-01411-4

References

  1. Brown, T. B. et al. Preprint at https://arxiv.org/abs/2005.14165 (2020).

  2. Yao, S. et al. Preprint at https://arxiv.org/abs/2210.03629 (2022).

  3. Wei, J. et al. Preprint at https://arxiv.org/abs/2201.11903 (2022).

  4. Schick, T. et al. Preprint at https://arxiv.org/abs/2302.04761 (2023).

Download references

Competing Interests

The author declares no competing interests.

Subjects

Latest on:

Nature Careers

Jobs

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing

Search

Advanced search

Quick links