2023-07-20 · auto-analysis (draft)

The Most Likely Outcomes of an AI Future with Emad Mostaque | EP #55

This chunk discusses the impact of AI on various industries, including journalism, healthcare, and knowledge work, and predicts widespread adoption of AI by next year. The podcast discusses the potential of AI in translating and customizing messages to better persuade individuals, the impact of media on mental health and longevity, the role of AI in education, and the challenges and opportunities of social media in shaping communities. The podcast discusses the importance of mental infrastructure, information diet, cognitive biases, and the potential impact of emerging technology on society over the next decade. The conversation discusses the impact of AI on the media industry, including the potential for AI-generated content, changes in business models, and the future of filmmaking. Emad Mostaque discusses the future impact of AI on various fields including education, healthcare, and me

▶ Watch on YouTube

Topics

AI in Journalism Media|AI|Truth ▶ 3:39

AI is impacting journalism by potentially enhancing truth and analysis but also creating challenges like clickbait and fake news.

AI in Healthcare Health|AI ▶ 0:23

Doctors and insurance premiums are becoming AI-assisted, transforming the knowledge sector.

AI in Knowledge Work AI|Work ▶ 2:25

By next year, every company with knowledge work will implement AI at scale, changing how tasks are performed.

AI in Technology Adoption AI|Adoption ▶ 1:46

AI technology is not yet widely adopted, with only 17% of US users having used chatGPT, but is expected to become more prevalent by next year.

AI and Contextual Persuasion AI|Compute ▶ 9:10

GPT models can translate points of view to better resonate with different contexts, potentially influencing persuasion.

Mental Health and Media Health|Economy|Geopolitics ▶ 12:00

Exposure to negative news can affect one's mindset and mental health, while positive media can have a positive impact on longevity and overall well-being.

Education and AI Education|Tech ▶ 14:05

AI can play a role in education, potentially shaping the mindset of children and their future outlook. The quality of education needs to be reconsidered for the future.

Social Media and Community Economy|Geopolitics|Health ▶ 16:05

Social media can both bring communities together and reinforce negative behaviors, highlighting the need for customization and positive reinforcement in the future.

Mental Infrastructure AI|Economy|Health ▶ 18:10

Emphasizes the importance of mental infrastructure and information diet, comparing it to physical diet and health.

Cognitive Biases and AI AI|Economy|Health ▶ 21:02

Discusses cognitive biases and the potential for AI to address them, highlighting the importance of awareness and context.

AI in Media AI|Media ▶ 27:02

Advances in AI have disrupted the media industry, reducing production costs and potentially leading to a democratization of content creation.

Future of Content Consumption AI|Media ▶ 29:39

The future of content consumption may involve personalized AI-generated content, with distribution methods changing due to AI technology.

Impact on Entertainment Industry AI|Media ▶ 34:43

Decisions and policies made now will significantly impact the entertainment industry for decades, with AI posing challenges and opportunities for actors and content creators.

AI in Filmmaking AI|Media ▶ 36:23

AI is transforming the nature of filmmaking, allowing for the creation of digital characters and scenes, potentially leading to a shift in the role of human actors.

Predictions made

open Unknown: By next year, every company with knowledge work will implement AI at scale.

EP #? · · due: next year · ▶ watch

“every company everywhere that has anything to do with knowledge work will implement this at scale and that's a crazy thought”

Your call:

open Unknown: AI technology will become more prevalent by next year.

EP #? · · due: next year · ▶ watch

“by next year that happens and I think every company everywhere that has anything to do with knowledge work will implement this at scale”

Your call:

open Emad Mostaque: AI will shape the mindset of children as they grow up, with significant plasticity in their minds.

EP #? · · due: unspecified · ▶ watch

“how you think impacts everything and how you think is to a large degree going to be shaped by the media and AI is going to shape that so it's a powerful lever that we all need to be paying attention to.”

Your call:

open Emad Mostaque: The new more powerful personalizable technology will either be controlled or control everybody.

EP #? · · due: unspecified · ▶ watch

“it's either controlled or control everybody”

Your call:

open Emad Mostaque: AI-generated content will be available in three to five years.

EP #? · · due: 3-5 years · ▶ watch

“I'd say probably three to five years so in that in that case it's not going to be cheap but it'll be there and then it'll get cheaper and cheaper in that case there is no distribution needed a person calls up whatever they want”

Your call:

open Emad Mostaque: The decisions and policies made now will significantly impact the entertainment industry for decades.

EP #? · · due: unspecified · ▶ watch

“the decisions the policies that we create today are going to take us down One path or another for the decades to come yeah it will affect the whole of Hollywood what's decided In This Moment here because there won't be another renegotiation for a while”

Your call:

open Emad Mostaque: Regulators will lose competitiveness to other countries and regulatory arbitrage will occur.

EP #? · · due: unspecified · ▶ watch

“you'll have regulatory Arbitrage yes and uh that is something across every industry”

Your call:

open Emad Mostaque: AI will be used in coding workflows in the next 2-5 years.

EP #? · · due: 2-5 years · ▶ watch

“there will be no code that doesn't use AI as part of their workflow”

Your call:

open Emad Mostaque: There will be a billion coders in the future as barriers to creating programs disappear.

EP #? · · due: unspecified · ▶ watch

“you can imagine there can be a billion coders in the future yeah”

Your call:

open Emad Mostaque: AI will revolutionize education, healthcare, and media by 2030.

EP #? · · due: 2030 · ▶ watch

“I can barely talk about 2030. it's everything everywhere all at once lots of s-curves acceleration these are inevitable now.”

Your call:

open Emad Mostaque: Every child will have their own AI assistant by 2033.

EP #? · · due: 2033 · ▶ watch

“every child will have their own AI that's learned for at least five years about them yes that can fetch them any information in any format of any type and write anything or create any video or movie for them to”

Your call:

open Emad Mostaque: A trillion dollars will be invested in the AI market, including self-driving cars.

EP #? · · due: trillion dollars going into the market · ▶ watch

“the capabilities will ramp up from here and so when I look at it I look at what the drivers are of why now it's first of all computation right and videos done an incredible job yeah right with their a100 and computation is continuing on Moore's Law it's not slowing down it's continuing to increase year on year a little bit exponential right well that is exponential well I mean yes it I I'm saying it's continuing to double on a regular basis yeah um what was considered Moore's Law and people have said oh it's going to eventually fall off as an s-curve well we're extending it and and for the next at least near-term future it's not slowing down so I think this is a very interesting thing for people to understand you had Moore's Law and again it was doubling and this was an individual chip what we do with these models is that we stick together thousands tens of thousands of these chips like how many a100s right now is stability using always about seven eight thousand by next year we will have 70 000 equivalent wow but what used to happen is as you stuck the chips together you ran a model so you take large amounts of data and you use these chips so I mean like we're using like maybe 10 megawatts 98 clean compared to the brain's 14 watts right 14 watts but then it compresses it down then it runs on 100 200 Watts or 25 watts actually okay put it down for some relevantage models so you do the pre-competition but the thing I see this the individual chips were doubling but what's the main breakthrough the last few years was is what happens when you stack them on top of each other to train a model you used to get to a hundred chips and then the performance collapsed because you couldn't move the data fast enough now you get to tens of thousands of chips and it keeps going up the performance of the model you don't have the big tail off anymore and so it's Moore's Law plus an additional scaling law and that's what enables these crazy performant models because you train longer you train bigger and then once the model is trained the end in the old internet the energy was used at the time of running the AI and then you'd collect the data that would be low energy relatively speaking it flips the equation because you pre-compute it you teach the curriculum up front and you send these little graduates out to the world such that you can have a language model now running on that MacBook yeah or an image model running on that MacBook drawing 25 to 35 watts of power to create a Renoir that can talk and recite Ulysses talking about Barbie you know that's insane all on your MacBook because we've done the pre-computation it's insane and this is because then what happens is the technology can spread when anyone can run it on their MacBook they”

Your call:

Numbers that matter

17% Only 17% of US users have used chatGPT despite its capabilities.
five to ten years The timeframe for the evolution of minds with the emergence of new technology.
70 billion The video game industry was a 70 billion dollar industry 10 years ago.
69 to 74 The average score on Metacritic went from 69 to 74 over a 10-year period.
40 billion to 50 billion Movies went from 40 billion to 50 billion in the last 10 years.
6.4 The average IMDb movie rating of the last 10 years has been 6.4 and has not changed.
500 Chinese film studios could release 500 products in every language we have technology now.
169 million images opted out of data sets for a company.
70 billion to 180 billion increase in the video game industry over the next so it's holidays.
7000 Number of medical articles written in journals every day.
30 billion dollars being put into the market self-driving cars at 100 billion dollars total this will be a trillion dollars going into this because do you know what I've got a trillion dollars 5G is this more important than 5G buy orders of magnitude orders of magnitude so it will get a trillion dollars going into it and the capabilities will ramp up from here and so when I look at it I look at what the drivers are of why now it's first of all computation right and videos done an incredible job yeah right with their a100 and computation is continuing on Moore's Law it's not slowing down it's continuing to increase year on year a little bit exponential right well that is exponential well I mean yes it I I'm saying it's continuing to double on a regular basis yeah um what was considered Moore's Law and people have said oh it's going to eventually fall off as an s-curve well we're extending it and and for the next at least near-term future it's not slowing down so I think this is a very interesting thing for people to understand you had Moore's Law and again it was doubling and this was an individual chip what we do with these models is that we stick together thousands tens of thousands of these chips like how many a100s right now is stability using always about seven eight thousand by next year we will have 70 000 equivalent wow but what used to happen is as you stuck the chips together you ran a model so you take large amounts of data and you use these chips so I mean like we're using like maybe 10 megawatts 98 clean compared to the brain's 14 watts right 14 watts but then it compresses it down then it runs on 100 200 Watts or 25 watts actually okay put it down for some relevantage models so you do the pre-competition but the thing I see this the individual chips were doubling but what's the main breakthrough the last few years was is what happens when you stack them on top of each other to train a model you used to get to a hundred chips and then the performance collapsed because you couldn't move the data fast enough now you get to tens of thousands of chips and it keeps going up the performance of the model you don't have the big tail off anymore and so it's Moore's Law plus an additional scaling law and that's what enables these crazy performant models because you train longer you train bigger and then once the model is trained the end in the old internet the energy was used at the time of running the AI and then you'd collect the data that would be low energy relatively speaking it flips the equation because you pre-compute it you teach the curriculum up front and you send these little graduates out to the world such that you can have a language model now running on that MacBook yeah or an image model running on that MacBook drawing 25 to 35 watts of power to create a Renoir that can talk and recite Ulysses talking about Barbie you know that's insane all on your MacBook because we've done the pre-computation it's insane and this is because then what happens is the technology can spread when anyone can run it on their MacBook they
8 billion Number of people on the planet.
50% Percentage of the world with cell phones.
5G and Starlink Global internet infrastructure.
2 to 10 years Timeline for serious concerns about AI.