High Output AI

Technology EN ↓ 42 episodes

A show dedicated to the latest news in the world of AI & Machine Learning. Distilling, reviewing and presenting the developments, breakthroughs and failures from the space. Hosted by Elliot & Tom, each week they bring a smattering of topics all with the purpose of helping you get the most from your application of AI.This is just a small part of their work in the space. Their channel contains how to's, explainers, tutorials and many more resources to help you become a better practitioner of AI. Check it out here https://www.youtube.com/channel/UCzuyWHAoUBXJBfVlDRJzbIA/featured

Author

High Output AI

Episodes

Amazon Buys One Medical | Podcast EP017 31.08.2022 55:43

Tom and Elliot talk about the recent acquisition of One Medical. Why Amazon would want it, where AI might fit into the picture and some tin hat theories of what their master plan entails. ⌛Timestamps 00:00 - Start 05:06 - AI For Marketing Misses the Mark 40:42 - AI For Grape Yield ⚡Links — Amazon official announcement https://press.aboutamazon.com/news-releases/news-release-details/amazon-and-on...

Do we need AI in Marketing and Agriculture? | EP016 27.07.2022 57:47

The cursed episode… This is V3 of this episode after a couple of tech failures… Tom and Elliot talk about two papers, one looking at AI in marketing and another in yield prediction for grapes. Both toe the line of "did we really need AI for this". ⌛Timestamps 00:00 - Start 05:06 - AI For Marketing Misses the Mark 40:42 - AI For Grape Yield ⚡Links — Predicting consumers’ choices in the age of the...

DALLE2 Pricing Model and Human Induced AI Bias in Language Models | EP015 19.07.2022 57:44

Tom and Elliot talk about DALLE-2’s new pricing model, and explore the topic of not only introducing data biases but also reasoning biases. ⌛Timestamps 00:00 - Start 02:10 - Topic 1 - DALLE2 Cost and Bias 28:31 - Topic 2 - Human Bias translates to AI Bias ⚡Links — DALLE-2 Pricing https://openai.com/api/pricing/ — AI Shows human like content-effects on reasoning https://arxiv.org/pdf/2207.07051.p...

Steve Jobs on Intelligence and AI Explainability | EP014 13.07.2022 1:09:47

Tom and Elliot talk about a Steve Jobs speech and his thoughts on intelligence and whether explainability will save us from stochastic parrots. ⌛Timestamps 00:00:00 - Start 00:06:00 - Topic 1 - Steve Jobs on Intelligence 00:36:24 - Topic 2 - AI explainability ⚡Links — Steve Jobs - Speech to the Academy of Achievement June 1982 https://www.youtube.com/watch?v=ymbD_a-G1IQ — Duncker’s Candle Proble...

Legalities of Copilot & $1 trillion on health admin | EP013 06.07.2022 1:09:46

This week Tom and Elliot dig into the new Copilot feature from GitHub and discuss a recent report from McKInsey that shows how the US spends nearly $1 Trillion on admin in the healthcare sector. 00:00:00 - Start 00:02:13 - Topic 1 - Github Copilot 00:03:37 - What is Copilot 00:07:01 - What's causing a push back 00:09:25 - What is Microsoft's plan here 00:11:16 - Does Copilot copy paste code? 00:12...

PARTI AI and Google's Investment in Healthcare | Podcast EP012 29.06.2022 1:10:01

Welcome to another episode. This week Tom and Elliot dig into the new PARTI image generation model from Google as well as Google's crazy investment numbers in health. Thanks for listening. 00:00:00 - Start 00:01:17 - What are we drinking? 00:04:22 - Alphabet's spend in Healthcare 00:13:12 - Searchable Patient Records 00:31:42 - Parti by Google 00:46:16 - Limitations of this model and others 01:06:...

Is LaMDA Sentient and Hospitality in Healthcare | EP011 22.06.2022 56:18

Welcome to another episode. This week Tom and Elliot dig into whether LaMDA is really sentient and how Healthcare can learn some lessons from hospitality. 00:00 - Start 02:40 - Is the AI Sentient? 03:13 - What is LaMDA 04:01 - What is Sentience? 05:08 - My thoughts: No 06:29 - What's going on then? 09:37 - Is the Turing Test a good test? 12:36 - How did people react 18:46 - Why hospitality matters...

Live Episode | EP010 15.06.2022 1:02:29

Tom and Elliot chat about a few topics in our first ever live episode. Let's hope two tech founders can work out a live stream.

DALLE-2 "Secret" Language & Did we forget how to build Nuclear? | EP009 08.06.2022 44:40

Tom and Elliot discuss the supposed secret language of DALLE-2 and some signs that the western world might have forgotten how to build new nuclear plants. TIMESTAMPS 00:00 Intro 04:00 Does DALLE-2 have a secret language? 06:06 What is OpenAI? 09:27 Failure Cases 13:04 Is this like Facebook's Failed Experiment? 17:06 Topic 2 Did we forget how to build power plants? 20:31 France's Nuclear Power 24:0...

Bizzare Signs of Recession and CRISPR Tomatoes | EP008 01.06.2022 1:06:49

Tom and Elliot discuss some of the more questionable signs that we might be in a recession and new work increasing Vitamin D in tomatoes using genetic modification with CRISPR. TIMESTAMPS 00:00:00 Intro 00:02:55 This Week's Topics 00:06:54 Topic 1 Bizzare indicators of recession 00:12:58 Hemline Index 00:18:55 R-Word Index 00:22:08 Men's Underwear Index 00:31:42 Bonus Index List 00:40:05 Stripper...

DeepMind's GATO model and the speed of human progress | EP007 23.05.2022 1:03:11

Tom and Elliot discuss Deepmind's new GATO model and an article looking at the speed of human progress and whether this century will be history's most important.   TIMESTAMPS 00:00 Intro 01:53 What are we drinking   05:42 Topic 1 Deepmind's GATO   29:47 Topic 2: This Can't Go On   58:51 Wrapping Up    ✍️SHOW NOTES Part 1 - DeepMind's GATO model -- DeepMind's GATO mode...

Bad AI metrics and How to understand things | EP006 10.05.2022 1:04:48

Tom and Elliot discuss how AI metrics get used and abused and a paper looking at how to understand things.   Part 1 -   How to Understand Things by Nabeel - https://nabeelqu.co/understanding -   Not Overthinking podcast episode that discusses Nabeel's article - https://notoverthinking.com/how-to-understand-things-2/ -   Visual image of "Bag of words" an...

Letting Ideas Incubate and The Morals of Online Prescriptions | EP005 10.05.2022 1:05:22

Tom and Elliot discuss a meta-analysis of how to leave time between problems to help solve them and a DEA investigation into online prescription services for ADHD.  Links Mentioned  - Does Incubation Enhance Problem Solving? A Meta-Analytic Review - https://www.researchgate.net/publication/24001208_Does_Incubation_Enhance_Problem_Solving_A_Meta-Analytic_Review - Plan, Do, Learn - My admi...

The Science of Problem Solving plus Elon open sourcing "The Algorithm" | EP004 02.05.2022 1:05:22

Tom and Elliot discuss the absurdity of open-sourcing Twitter's timeline algorithm and what we know about the science of solving problems.   Links Mentioned:  - https://rs.io/the-science-of-problem-solving/ - https://tomkwhite.com - https://elliotcsmith.com  #podcast #tech #ai #startup

Defeating nations with grass and counterfactual value | EP003 18.04.2022 45:26

Tom and Elliot discuss how grass helped horseback tribes conquer nations and different ways to look at value.    Links Mentioned:  - https://perell.com/note/the-counterfactual-theory-of-value/ - https://tomkwhite.com - https://elliotcsmith.com

AI Takeoff Speeds and Hammock Driven Development | EP002 11.04.2022 58:10

Tom and Elliot discuss how quickly AI might become super intelligent and how to get your best work done by spending time in a hammock.   Links Mentioned:  - https://astralcodexten.substack.com/p/yudkowsky-contra-christiano-on-ai?s=r- https://en.wikipedia.org/wiki/GPT-3 - lesswrong.com/  - http://gwern.net/    - Hammock Driven Development: https://www.youtube.com/watch?v=f8...

Mic Check One Two | EP001 03.04.2022 1:01:00

Tom and Elliot discuss Linkedin DeepFakes and the Kano Model for classifying features.   Apologies for the echo, we'll get that sorted for the next episode.   Links Mentioned:  - https://www.npr.org/2022/03/27/1088140809/fake-linkedin-profiles  - https://www.productplan.com/glossary/kano-model/  - https://twitter.com/elliot_c_smith - https://twitter.com/tom_k_white  -...