brainsteam.co.uk/brainsteam/content/posts/2023/12/20/NLP in the Post-LLM World.md

---
categories:
- AI and Machine Learning
date: '2023-12-20 06:37:14'
draft: false
tags:
- nlp
title: NLP in the Post-LLM World
type: posts
url: /2023/12/20/nlp-in-the-post-llm-world/
---

<!-- wp:paragraph -->
<p>I really enjoyed diving into Seb Ruder's <a href="https://nlpnewsletter.substack.com/p/nlp-research-in-the-era-of-llms">latest NLP Newsletter</a> which focuses on all the areas of NLP that are still in desperate need of attention in a post-LLM world.</p>
<!-- /wp:paragraph -->

<!-- wp:quote -->
<blockquote class="wp-block-quote"><!-- wp:paragraph -->
<p>In an era where running state-of-the-art models requires a garrison of expensive GPUs, what research is left for academics, PhD students, and newcomers to NLP without such deep pockets?</p>
<!-- /wp:paragraph -->

<!-- wp:paragraph -->
<p>...while massive compute often achieves breakthrough results, its usage is often inefficient. Over time, improved hardware, new techniques, and novel insights provide opportunities for dramatic compute reduction...</p>
<!-- /wp:paragraph --></blockquote>
<!-- /wp:quote -->

<!-- wp:paragraph -->
<p> I wrote about some of the same issues in my post <a href="https://brainsteam.co.uk/2023/03/25/nlp-is-more-than-just-llms/">NLP is more than just LLMs</a> earlier this year and <a href="https://brainsteam.co.uk/2023/12/05/carbon-footprint-of-generative-models/">I recently speculated</a> about how current industry AI darlings, sexy-scale-up companies very much in "growth" mode as opposed to incumbents in "cost-saving" mode, are just not incentivised to be compute-efficient.</p>
<!-- /wp:paragraph -->

<!-- wp:paragraph -->
<p>If you are just starting out in this space there are plenty of opportunities and lots of problems to solve - particularly around trust, reliability and energy efficiency.</p>
<!-- /wp:paragraph -->