<description>&lt;p&gt;Gwern is a pseudonymous researcher and writer. He was one of the first people to see LLM scaling coming. If you've read his &lt;a target="_blank" href="https://gwern.net/"&gt;blog&lt;/a&gt;, you know he's one of the most interesting polymathic thinkers alive.&lt;/p&gt;&lt;p&gt;In order to protect Gwern's anonymity, I proposed interviewing him in person, and having my friend &lt;a target="_blank" href="https://x.com/ChrisPainterYup"&gt;Chris Painter&lt;/a&gt; voice over his words after. This amused him enough that he agreed.&lt;/p&gt;&lt;p&gt;After the episode, I convinced Gwern to create a donation page where people can help sustain what he's up to. Please go &lt;a target="_blank" href="https://donate.stripe.com/6oE9DTgaf6oD0M03cc"&gt;here&lt;/a&gt; to contribute.&lt;/p&gt;&lt;p&gt;Read the full transcript &lt;a target="_blank" href="https://www.dwarkeshpatel.com/p/gwern-branwen"&gt;here&lt;/a&gt;.&lt;/p&gt;&lt;p&gt;Sponsors:&lt;/p&gt;&lt;p&gt;* &lt;a target="_blank" href="https://jane-st.co/dwarkesh"&gt;Jane Street&lt;/a&gt; is looking to hire their next generation of leaders. Their deep learning team is looking for ML researchers, FPGA programmers, and CUDA programmers. Summer internships are open - if you want to stand out, take a crack at their new Kaggle competition. To learn more, go here: &lt;a target="_blank" href="https://jane-st.co/dwarkesh"&gt;https://jane-st.co/dwarkesh&lt;/a&gt;&lt;/p&gt;&lt;p&gt;* Turing provides complete post-training services for leading AI labs like OpenAI, Anthropic, Meta, and Gemini. They specialize in model evaluation, SFT, RLHF, and DPO to enhance models’ reasoning, coding, and multimodal capabilities. Learn more at&lt;a target="_blank" href="https://turing.com/dwarkesh"&gt; &lt;/a&gt;&lt;a target="_blank" href="http://turing.com/dwarkesh"&gt;turing.com/dwarkesh&lt;/a&gt;.&lt;/p&gt;&lt;p&gt;* This episode is brought to you by &lt;a target="_blank" href="https://stripe.com/"&gt;Stripe&lt;/a&gt;, financial infrastructure for the internet. Millions of companies from Anthropic to Amazon use Stripe to accept payments, automate financial processes and grow their revenue.&lt;/p&gt;&lt;p&gt;If you’re interested in advertising on the podcast, check out &lt;a target="_blank" href="https://www.dwarkeshpatel.com/p/advertise"&gt;this page&lt;/a&gt;.&lt;/p&gt;&lt;p&gt;Timestamps&lt;/p&gt;&lt;p&gt;00:00:00 - Anonymity&lt;/p&gt;&lt;p&gt;00:01:09 - Automating Steve Jobs&lt;/p&gt;&lt;p&gt;00:04:38 - Isaac Newton's theory of progress&lt;/p&gt;&lt;p&gt;00:06:36 - Grand theory of intelligence&lt;/p&gt;&lt;p&gt;00:10:39 - Seeing scaling early&lt;/p&gt;&lt;p&gt;00:21:04 - AGI Timelines&lt;/p&gt;&lt;p&gt;00:22:54 - What to do in remaining 3 years until AGI&lt;/p&gt;&lt;p&gt;00:26:29 - Influencing the shoggoth with writing&lt;/p&gt;&lt;p&gt;00:30:50 - Human vs artificial intelligence&lt;/p&gt;&lt;p&gt;00:33:52 - Rabbit holes&lt;/p&gt;&lt;p&gt;00:38:48 - Hearing impairment&lt;/p&gt;&lt;p&gt;00:43:00 - Wikipedia editing&lt;/p&gt;&lt;p&gt;00:47:43 - Gwern.net&lt;/p&gt;&lt;p&gt;00:50:20 - Counterfactual careers&lt;/p&gt;&lt;p&gt;00:54:30 - Borges &amp; literature&lt;/p&gt;&lt;p&gt;01:01:32 - Gwern's intelligence and process&lt;/p&gt;&lt;p&gt;01:11:03 - A day in the life of Gwern&lt;/p&gt;&lt;p&gt;01:19:16 - Gwern's finances&lt;/p&gt;&lt;p&gt;01:25:05 - The diversity of AI minds&lt;/p&gt;&lt;p&gt;01:27:24 - GLP drugs and obesity&lt;/p&gt;&lt;p&gt;01:31:08 - Drug experimentation&lt;/p&gt;&lt;p&gt;01:33:40 - Parasocial relationships&lt;/p&gt;&lt;p&gt;01:35:23 - Open rabbit holes&lt;/p&gt; &lt;br/&gt;&lt;br/&gt;Get full access to Dwarkesh Podcast at &lt;a href="https://www.dwarkeshpatel.com/subscribe?utm_medium=podcast&amp;#38;utm_campaign=CTA_4"&gt;www.dwarkeshpatel.com/subscribe&lt;/a&gt;</description>

Dwarkesh Podcast

Dwarkesh Patel

Gwern Branwen - How an Anonymous Researcher Predicted AI's Trajectory

NOV 13, 202496 MIN
Dwarkesh Podcast

Gwern Branwen - How an Anonymous Researcher Predicted AI's Trajectory

NOV 13, 202496 MIN

Description

<p>Gwern is a pseudonymous researcher and writer. He was one of the first people to see LLM scaling coming. If you've read his <a target="_blank" href="https://gwern.net/">blog</a>, you know he's one of the most interesting polymathic thinkers alive.</p><p>In order to protect Gwern's anonymity, I proposed interviewing him in person, and having my friend <a target="_blank" href="https://x.com/ChrisPainterYup">Chris Painter</a> voice over his words after. This amused him enough that he agreed.</p><p>After the episode, I convinced Gwern to create a donation page where people can help sustain what he's up to. Please go <a target="_blank" href="https://donate.stripe.com/6oE9DTgaf6oD0M03cc">here</a> to contribute.</p><p>Read the full transcript <a target="_blank" href="https://www.dwarkeshpatel.com/p/gwern-branwen">here</a>.</p><p>Sponsors:</p><p>* <a target="_blank" href="https://jane-st.co/dwarkesh">Jane Street</a> is looking to hire their next generation of leaders. Their deep learning team is looking for ML researchers, FPGA programmers, and CUDA programmers. Summer internships are open - if you want to stand out, take a crack at their new Kaggle competition. To learn more, go here: <a target="_blank" href="https://jane-st.co/dwarkesh">https://jane-st.co/dwarkesh</a></p><p>* Turing provides complete post-training services for leading AI labs like OpenAI, Anthropic, Meta, and Gemini. They specialize in model evaluation, SFT, RLHF, and DPO to enhance models’ reasoning, coding, and multimodal capabilities. Learn more at<a target="_blank" href="https://turing.com/dwarkesh"> </a><a target="_blank" href="http://turing.com/dwarkesh">turing.com/dwarkesh</a>.</p><p>* This episode is brought to you by <a target="_blank" href="https://stripe.com/">Stripe</a>, financial infrastructure for the internet. Millions of companies from Anthropic to Amazon use Stripe to accept payments, automate financial processes and grow their revenue.</p><p>If you’re interested in advertising on the podcast, check out <a target="_blank" href="https://www.dwarkeshpatel.com/p/advertise">this page</a>.</p><p>Timestamps</p><p>00:00:00 - Anonymity</p><p>00:01:09 - Automating Steve Jobs</p><p>00:04:38 - Isaac Newton's theory of progress</p><p>00:06:36 - Grand theory of intelligence</p><p>00:10:39 - Seeing scaling early</p><p>00:21:04 - AGI Timelines</p><p>00:22:54 - What to do in remaining 3 years until AGI</p><p>00:26:29 - Influencing the shoggoth with writing</p><p>00:30:50 - Human vs artificial intelligence</p><p>00:33:52 - Rabbit holes</p><p>00:38:48 - Hearing impairment</p><p>00:43:00 - Wikipedia editing</p><p>00:47:43 - Gwern.net</p><p>00:50:20 - Counterfactual careers</p><p>00:54:30 - Borges & literature</p><p>01:01:32 - Gwern's intelligence and process</p><p>01:11:03 - A day in the life of Gwern</p><p>01:19:16 - Gwern's finances</p><p>01:25:05 - The diversity of AI minds</p><p>01:27:24 - GLP drugs and obesity</p><p>01:31:08 - Drug experimentation</p><p>01:33:40 - Parasocial relationships</p><p>01:35:23 - Open rabbit holes</p> <br/><br/>Get full access to Dwarkesh Podcast at <a href="https://www.dwarkeshpatel.com/subscribe?utm_medium=podcast&#38;utm_campaign=CTA_4">www.dwarkeshpatel.com/subscribe</a>