AI Predictions ― 2024 to 2030 ― Year By Year Breakdown w/ Insider Info

David Shapiro
28 Jul 202426:33

Summary

TLDRThe speaker revisits AI predictions, acknowledging past inconsistencies and offering an updated outlook on AI's trajectory to 2030. They discuss anticipated milestones like the release of GPT 5 and commercial robots, the potential for AGI by 2027, and the broader impacts on society, including job displacement and new economic paradigms. The script suggests a period of disillusionment before breakthroughs lead to a 'New Renaissance' by 2029, with AI and robotics becoming integral to daily life.

Takeaways

  • 🔮 The speaker anticipates a 'down term' for AI, with a period of disillusionment ahead but maintains a long-term optimistic view.
  • 🤖 There is an expectation for incremental improvements in AI models, with GPT 5 and Claude 4 expected to be released around late 2024 or early 2025.
  • 📉 The speaker notes a slowdown in AI breakthroughs due to increasing costs, but an acceleration in the release of foundation models.
  • 🧩 The multimodality aspect of AI is set to expand, with more AI demonstrations from tech giants, despite some being exaggerated or misleading.
  • 🤖🏭 The release of commercial and domestic robots is expected to be a significant development in the AI and robotics space.
  • 🏢 For 2025, the speaker predicts a year of disillusionment as AI models like GPT 5 and Claude 4 may not reach the AGI level as hoped.
  • 📊 AI models are expected to reach the 95th percentile across multiple benchmarks, which traditionally signals a 'solved problem' in machine learning.
  • 💼 2025 may see more enterprise deployment of AI, with small and medium businesses leading the adoption curve due to their agility.
  • 📈 By 2026, AI models are predicted to be 'Enterprise ready' and considered the first true general-purpose models, potentially qualifying as AGI.
  • 🚀 The year 2029 is envisioned as a turning point, with the onset of new technologies like commercial nuclear fusion, quantum computing, and advanced AI leading to a 'New Renaissance'.
  • ⏳ The speaker expects societal and economic shifts by 2030, possibly introducing concepts like post-labor economics and the potential for longevity escape velocity.

Q & A

  • What is the main topic of the video script?

    -The main topic of the video script is the prediction of future milestones in the field of artificial intelligence, robotics, and their impact on society and the economy from the present until 2030.

  • What does the speaker believe about the current state of AI and its future developments?

    -The speaker believes that although AI is entering a period of disillusionment and slower progress, they remain optimistic about its future. They predict incremental improvements and multimodal capabilities in AI, along with the release of more commercial and domestic robots.

  • What is the expected timeline for the release of GPT 5 and Claude 4 according to the script?

    -GPT 5 is expected to be released in late 2024 or early 2025, and Claude 4 is also expected around the same time.

  • What does the speaker mean by 'diminishing returns' in the context of AI models?

    -The speaker refers to the point where further advancements in AI models become more expensive and less impactful, suggesting that the pace of breakthroughs will slow down even if the release of models becomes more frequent.

  • What is the speaker's view on the current benchmarks for measuring AI intelligence?

    -The speaker is critical of the current benchmarks, stating that they are not particularly useful or helpful for measuring real intelligence, as they do not account for long-term horizons, chaotic environments, adaptation, or real-world application.

  • What are the speaker's predictions for the year 2025 in terms of AI adoption and impact?

    -The speaker predicts that 2025 will be a year of disillusionment, with AI models like GPT 5 and Claude 4 not reaching AGI but showing significant improvements. They also expect more enterprise deployment of AI tools by small and medium businesses.

  • What does the speaker anticipate for the year 2026 in the AI industry?

    -The speaker anticipates that 2026 will be the year when AI models are considered enterprise-ready, with general-purpose models that can be applied across various industries and modalities.

  • What is the speaker's perspective on the geopolitical situation and its relation to AI by 2028?

    -The speaker suggests that 2028 might be a critical year with a potential new arms race and conflict with China, coinciding with the mass integration of AGI and robotics, which could lead to mass layoffs and become a significant political issue.

  • What does the speaker envision for the year 2029 in terms of technological advancements?

    -The speaker envisions 2029 as the beginning of a new renaissance with the mainstream adoption of quantum computing, commercial nuclear fusion reactors, and the widespread impact of AI on various fields such as material science and genetic engineering.

  • What is the speaker's long-term outlook for the year 2030 and beyond?

    -The speaker's long-term outlook for 2030 and beyond is optimistic, predicting the start of a new golden era characterized by the intelligence age, where AI and other technologies will reshape the economy, society, and geopolitics, potentially leading to a new paradigm of post-labor economics.

  • What does the speaker suggest about the potential societal and economic changes due to AI and automation?

    -The speaker suggests that AI and automation will lead to significant societal and economic changes, including the need for new economic paradigms such as post-labor economics, and the potential for universal basic income (UBI), while also highlighting the challenges of finding meaning and purpose in a post-labor society.

Outlines

00:00

🤖 AI Predictions and the Trough of Disillusionment

The speaker revisits their AI predictions, acknowledging a recent shift in their stance due to new insights from industry insiders. They clarify that while they expect a period of disappointment with AI, they remain optimistic. The summary of expected milestones from 2024 to 2030 is based on these insider perspectives. The speaker predicts incremental improvements in AI models like GPT 5 and Claude 4, with a focus on cost and the pace of breakthroughs, and hints at the potential for commercial and domestic robots to make significant impacts by the end of 2024.

05:02

🔮 Anticipating the Future of AI and its Enterprise Integration

The speaker forecasts that AI models will reach the 95th percentile across various benchmarks by 2025, which traditionally signals a 'solved problem' in machine learning. They discuss the limitations of current benchmarks and the need for new ones that better measure real-world intelligence. The expectation is that refined, cost-effective AI models will lead to increased enterprise deployment, particularly among small and medium businesses. However, there's skepticism among enterprise corporations about AI's readiness for large-scale integration, suggesting a slower adoption rate at the enterprise level.

10:03

🏢 The Path to Enterprise Readiness for AI Models

The speaker anticipates that by 2026, AI models will be considered enterprise-ready, following a period of disillusionment. They envision models like Claude 5 or GPT 6 as the first true general-purpose AI models, applicable across various industries and modalities. The speaker also speculates that by this time, we might be close to achieving AGI (Artificial General Intelligence), which would significantly impact the job market and the economy, possibly leading to discussions around AI safety and job protection measures.

15:04

🎬 AI's Role in the Creation of Hollywood Blockbusters

The speaker predicts that by 2027, we could see the first feature-length Hollywood blockbuster film created entirely by AI, marking a significant milestone in the industry's use of AI. They also anticipate broader impacts of AI, including potential job displacement and political issues related to AI safety and economic policies like universal basic income. The geopolitical landscape is also expected to shift, with countries accelerating their development of AI technologies.

20:05

🚀 The Convergence of Technologies in 2029: A New Renaissance

The speaker foresees 2029 as a pivotal year when data centers, commercial nuclear fusion reactors, quantum computing, and other technologies will become mainstream, leading to a new renaissance. They predict significant advancements in material science, genetic engineering, and medical breakthroughs, contributing to a compounding effect of technological progress. This period is expected to bring about a new level of optimism and a shift in the geopolitical and economic landscape.

25:06

🌐 The Socioeconomic Shifts and the Dawn of the Intelligence Age

The speaker anticipates that by 2030, we will have entered a new era characterized by the widespread adoption of AI, which they refer to as the 'Intelligence Age'. They predict that this period will see the establishment of new economic paradigms, such as post-labor economics, and the potential realization of longevity escape velocity. The speaker also expresses hope for a new approach to finding meaning in life and reshaping the geopolitical conversation, envisioning a future of abundance and optimism.

🕊️ Hopes for a Peaceful Transition to a Technologically Advanced Future

In the final paragraph, the speaker reflects on the potential challenges and opportunities ahead, emphasizing the importance of avoiding conflict and focusing on healing the planet. They express hope for a peaceful transition to a future where advanced technologies like nuclear fusion and AGI contribute to a better world. The speaker also acknowledges the long-term nature of geopolitical healing and the need for new paradigms to address the shifts brought about by rapid technological advancements.

Mindmap

Keywords

💡Artificial Intelligence (AI)

Artificial Intelligence refers to the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions. In the video's theme, AI is central to the discussion of technological advancement and its impact on society. The script mentions AI's potential and current limitations, indicating a period of 'disillusionment' due to unmet expectations, and also references AI models like GPT and Claude, which are expected to evolve significantly by 2030.

💡Disillusionment

Disillusionment in the context of the video refers to a period of disappointment or the loss of confidence in the progress of AI. The script predicts a 'down term' in AI development where expectations for rapid advancements may not be met, leading to a sense of disillusionment among the public and industry insiders alike.

💡Industry Insiders

Industry insiders are individuals who have deep knowledge and experience in a particular field or industry. The script mentions that the speaker has been in conversation with such insiders from around the world to gain a more accurate perspective on the future of AI, indicating the importance of expert opinion in predicting industry trends.

💡General Purpose Models

General purpose models in AI are versatile and can be applied across various domains and tasks without needing significant modifications. The script discusses the anticipation of such models becoming 'Enterprise ready' by 2026, suggesting that these models will be capable of handling a wide range of applications, making them valuable for businesses.

💡Enterprise Adoption

Enterprise adoption refers to the process by which large corporations begin to implement and integrate new technologies into their operations. The video script predicts that by 2026, AI models will be considered ready for enterprise adoption, indicating a significant milestone in the mainstream integration of AI into various industries.

💡Benchmarks

In the context of AI, benchmarks are standardized tests or metrics used to evaluate the performance of AI models. The script critiques current benchmarks as not being particularly useful for measuring true intelligence, and predicts that as AI models improve, existing benchmarks will become obsolete, necessitating the development of new ones.

💡Cost-Effectiveness

Cost-effectiveness pertains to the balance between the cost of a product or service and the benefits it provides. The script suggests that as AI models become more refined, they will also become more cost-effective, which could lead to increased adoption in various sectors, including small and medium businesses.

💡Robotics

Robotics is the branch of technology that deals with the design, construction, operation, and use of robots. The video script highlights robotics as a significant area of development, predicting that advancements in this field will bring about a 'step change' in AI and robotics, with commercial and domestic robots becoming more prevalent.

💡Post-Labor Economics

Post-labor economics refers to a theoretical economic system where the need for human labor is significantly reduced due to automation and technological advancements. The script anticipates a shift towards this system as AI and robotics advance, necessitating a rethinking of traditional economic models and social structures.

💡Longevity Escape Velocity

Longevity escape velocity is a concept that suggests a point in time when advancements in medical technology and healthcare will allow life expectancy to increase at a rate that outpaces the aging process. The script mentions this concept in relation to the potential impacts of AI and other technologies like quantum computing by 2030.

💡Intelligence Age

The Intelligence Age is a term used in the script to describe a future era characterized by the dominance of AI and intelligent systems in all aspects of life. It is presented as a successor to the Information Age and is expected to bring about significant changes in society, economy, and technology by 2030.

Highlights

Expectations of a down term and a period of disillusionment with artificial intelligence are discussed.

The speaker remains optimistic about AI despite acknowledging a disappointing period ahead.

Predictions for AI milestones from now until 2030 are based on insider industry conversations.

GPT 5 and Claude 4 are expected to be released in late 2024 or early 2025.

AI advancements are slowing down due to increasing costs, despite faster model release cadence.

Incremental changes in AI models are significant, especially in context windows and reasoning abilities.

Multimodality and AI demonstrations by tech giants are critiqued for overpromising and underdelivering.

The release of commercial and domestic robots is anticipated to be a significant development in AI.

Disney's advanced robotics program is highlighted for its potential beyond entertainment.

2025 is predicted as the year of disillusionment, with AI models not reaching AGI status.

AI models are expected to reach the 95th percentile across benchmarks by 2025.

The need for new benchmarks to measure true intelligence beyond current testing methods is emphasized.

Enterprise deployment of AI is expected to increase in 2025, with SMBs leading the charge.

Job displacement due to AI advancements is predicted to become more noticeable in 2025.

2026 is posited as the year when AI models will be considered enterprise-ready and general purpose.

AGI is anticipated in 2027, with varying confidence intervals discussed.

The geopolitical implications of AI advancements, particularly regarding China, are explored.

Potential societal and economic shifts due to AGI and robotics are predicted for 2028.

2029 is envisioned as the beginning of a new Renaissance marked by technological convergence.

The transition to a new economic paradigm, post-labor economics, is anticipated by 2030.

Longevity escape velocity and its societal impacts are predicted to be significant by 2030.

The speaker's hope for a new era of abundance and its potential to reshape geopolitics is expressed.

Transcripts

play00:00

so it's been a little while since I've

play00:01

done one of these prediction videos

play00:02

where I just kind of lay everything out

play00:04

and I realize that there's probably been

play00:06

a little bit of perceived inconsistency

play00:09

in my position lately so I wanted to

play00:11

take a moment to clear all that up and

play00:12

so what I mean is that you know like yes

play00:14

I am expecting a down term uh we're

play00:17

entering into the truff of

play00:18

disillusionment and I do think that

play00:20

we're in for a little bit of a

play00:21

disappointing uh period with artificial

play00:23

intelligence however I still remain

play00:26

pretty optimistic so let's just dive in

play00:28

and let's go over the next few years

play00:30

from now until 2030 kind of some of the

play00:33

Milestones that I expect to see and what

play00:35

I want to tell you before we dive in is

play00:37

that this is not just speculation on my

play00:40

part I have been talking to Industry

play00:42

insiders uh from around the world and uh

play00:45

from looking at AI from various

play00:48

perspectives um and conversations behind

play00:50

the scenes for the last couple weeks so

play00:53

I'm a little bit more oriented than I

play00:55

used to be so hopefully this uh will

play00:58

prove to be a little bit more accurate

play01:00

than some of my predictions in the past

play01:01

so without further Ado let's Dive Right

play01:03

In so first how are we going to uh close

play01:06

out 2024 now obviously in the past I was

play01:09

predicting AGI by the end of this year

play01:12

and um but you know with other people

play01:13

commenting uh you know such as Leopold

play01:16

Ashen Runner um it looks like 2027 is

play01:19

kind of when people are settling on AGI

play01:22

then the confidence intervals you know

play01:24

are give and take a little bit we'll get

play01:26

into that in just a minute but let's

play01:28

focus on the rest of this year so GPT 5

play01:31

is expected to drop uh this year as

play01:33

early as fall but likely December or

play01:36

early 2025 Claud 4 is um also expected

play01:40

around that time um now what anthropic

play01:43

is doing right now is they're working on

play01:45

the rest of the 3.5 family but the

play01:47

Cadence is accelerating with releasing

play01:49

these Foundation models um and while I

play01:52

have said that AI is slowing down what

play01:54

I'm mostly looking at is the cost uh

play01:56

increasing so basically the uh the break

play01:59

throughs that we that we're hoping to

play02:01

see are going to be happening slower

play02:03

even if the Cadence of model releases is

play02:06

a little bit faster cuz I was looking at

play02:08

the timelines and it was I think it was

play02:10

from from gpt3 to GPT 4 was a full three

play02:13

years and so GPT 4 to GPT 5 looks like

play02:17

it might be two two and a halfish years

play02:20

so the overall Cadence is more or less

play02:22

keeping up to Pace but the but there are

play02:25

more or less just incremental changes

play02:27

and so yes incremental changes can do a

play02:30

lot particularly the incremental changes

play02:32

we've seen on context Windows of

play02:34

reasoning abilities and those sorts of

play02:35

things we're also going to see a lot

play02:37

more M multimodality obviously um some

play02:40

of the demos that have been presented by

play02:41

Google and Microsoft and open AI um the

play02:44

demos were one thing some of them were

play02:45

faked um and in many cases they've over

play02:48

promised and underd delivered um but

play02:51

that's kind of the news that yall are

play02:52

all familiar with one thing that I think

play02:55

that we're that is probably going to be

play02:56

like the last you know big shocking

play02:58

thing that we see this year is is the

play03:00

release of more commercial and

play03:01

potentially domestic robots um if you

play03:04

haven't been keeping up with the news

play03:06

there are so many companies in China

play03:08

America and and other places that are

play03:11

working diligently on humanoid robots

play03:13

for various purposes Disney

play03:14

interestingly has some of the most

play03:16

advanced robotics programs but they're

play03:18

just using it for like Dunt doubles and

play03:20

animatronics and I'm like you guys have

play03:22

no idea what you're sitting on you have

play03:24

the most lifelike robots and you're

play03:25

using it for entertainment what are you

play03:27

doing anyways um I'll get off my Soap

play03:30

Box about Disney so uh I think that I

play03:33

think that you know cuz if you've seen

play03:34

like the Boston Dynamics videos and you

play03:37

and everything else coming out about

play03:39

humanoid robots I think that this is

play03:41

going to be the next step shift um The

play03:43

Next Step change in the AI and robotic

play03:46

space because again models we're we're

play03:49

getting to the point of diminishing

play03:50

returns with the current paradig of

play03:51

models meaning that getting those next

play03:53

breakthroughs is going to be

play03:55

exponentially more expensive but at the

play03:57

same time once you get to a certain

play03:58

threshold it doesn't really matter cuz

play04:00

like okay you can have like once you get

play04:02

models that are PhD smart what are you

play04:04

going to go after that like 10x PhD okay

play04:07

sure like yes we can debate intelligence

play04:09

till the cows come home um but my point

play04:12

here is that 2024 we're not going to

play04:14

have a whole lot more exciting news I

play04:16

think robotics is going to be the next

play04:17

most exciting thing um Sora and video

play04:20

generators are going to be really

play04:21

interesting um but I suspect that we're

play04:23

going to find that getting getting to

play04:25

like super crisp like Hollywood ready 4K

play04:28

and 8K video is going to be a little bit

play04:31

harder just in the same way that you

play04:33

know language models have mastered

play04:34

language but getting it to Shakespeare

play04:36

we're not there yet now 2025 I think is

play04:40

going to be the year of the

play04:41

disillusionment where a lot of people

play04:43

myself included are going to be kind of

play04:45

disappointed like okay GPT 5 hit and

play04:47

it's definitely smarter it's definitely

play04:49

PhD level in some respects but it's not

play04:53

quite going to be AGI same thing with

play04:55

Claude 4 um I think it's pretty safe to

play04:58

assume that GPT 5 and CL for will be out

play05:01

by 2025 or probably early 2025 now one

play05:05

thing that I will say is that um I

play05:07

expect these models to get to like the

play05:09

95th percentile across multiple

play05:11

benchmarks um and you might say like

play05:14

okay that sounds pretty good and in the

play05:16

old Paradigm of machine learning once

play05:18

you get to the 95th percentile um that's

play05:20

basically considered a solved problem in

play05:22

machine learning and so right now

play05:24

depending on which Benchmark you're

play05:25

looking at they you know anywhere from

play05:26

the 40th percentile to the 85th

play05:28

percentile and those last few percentage

play05:31

points usually are much much harder to

play05:33

to solve at least until you get an

play05:35

algorithmic breakthrough and this has

play05:37

happened many many times in the AI space

play05:40

um obviously before Transformers um but

play05:43

you know I remember listening to podcast

play05:46

about um you know uh like uh decision

play05:49

trees and those sorts of things and XG

play05:51

boost um then when those new techniques

play05:53

were introduced uh you know many many

play05:56

problems went from 70th percentile to

play05:58

95th percentile and then those

play06:00

competitions were no longer interesting

play06:02

after that so what I mean like when like

play06:06

what are the implications when we get

play06:07

you know to the 95th percentile on all

play06:09

the benchmarks today I think we're going

play06:11

to realize that a lot of the benchmarks

play06:12

to that exist today are not particularly

play06:14

useful or helpful which is why I've been

play06:16

kind of critical of them um is because

play06:19

it's like okay it's good at taking a

play06:20

test but real intelligence is about

play06:22

longtime Horizons chaotic environments

play06:25

um and adaptation and there's not really

play06:28

testing that right now so like

play06:30

when people ask me like what's a good

play06:31

Benchmark for intelligence I'm like

play06:33

there are none right now there are

play06:34

literally no good benchmarks out there

play06:36

that I have been impressed by that

play06:37

really measure intelligence yes some of

play06:40

them are really interesting um and and

play06:43

you know definitely are measuring some

play06:45

aspects of intelligence such as

play06:46

reasoning in an abstract state but you

play06:49

know you can have the best person who's

play06:51

you know great at math great at you know

play06:54

reasoning and Abstract uh Concepts who

play06:57

still has no Street smarts or real world

play06:59

elligence or the ability to make things

play07:01

happen in the real world um so first

play07:04

contact with the real world is going to

play07:06

be very messy for these AI models but

play07:09

what I will say is that as these

play07:11

benchmarks uh get met and also as we get

play07:14

more distilled models more uh refined

play07:16

models um that they're going to be much

play07:19

more cost- effective because it's like

play07:20

okay if you have a model that is

play07:22

basically free to run and it's as good

play07:24

as you know someone with two or three

play07:26

years of experience that's going to

play07:28

start turning some heads and start

play07:30

replacing some jobs so I do suspect that

play07:32

2025 we're going to see a lot more

play07:34

Enterprise deployment what I'm hearing

play07:37

out there right now is that most people

play07:38

deploying AI are small and medium

play07:40

businesses because they can pivot faster

play07:43

and I am hearing that Enterprise

play07:44

corporations are looking at this stuff

play07:47

but none of the leadership are convinced

play07:49

that it is Enterprise ready and we'll

play07:50

talk about that in just a second um

play07:52

however because the small and medium

play07:55

businesses are going to are going to

play07:56

start adopting more AI tools either tool

play07:59

you know like co-pilot provided by

play08:01

Microsoft and GitHub or tools built by

play08:03

startups you're going to see a surge in

play08:06

people basically either in the startup

play08:09

space building these new tools or at

play08:12

some of these larger companies um

play08:14

building AI products for companies and

play08:16

then of course you're going to have some

play08:18

internal hiring for uh uh for internal

play08:20

talent to deploy and utilize AI um but I

play08:24

think that 2025 is still going to be a

play08:26

relatively tepid year at least at the

play08:28

Enterprise scale

play08:30

um but smbs usually and SB if you're not

play08:32

familiar with small and medium

play08:33

businesses so those are those are like

play08:35

the mom and pop shops up to like you

play08:37

know 400 employees or so a th000

play08:40

employees I'm not really sure where the

play08:42

uh the the transition point from SMB to

play08:45

Enterprise is today and of course

play08:47

there's many intermediary steps anyways

play08:49

getting lost in the weeds my point is is

play08:51

that smaller companies can pivot faster

play08:53

so we're probably going to see more

play08:55

hiring in the startup space um and the

play08:58

SMB space where AI is concerned and then

play09:00

of course there's also going to be um a

play09:02

lot of internal hiring at the uh at like

play09:04

the Microsoft and Google and those sorts

play09:06

of things um but those companies tend to

play09:08

be a little bit more insular um meaning

play09:10

like if you're in the club you're in the

play09:12

club and they don't really like

play09:13

Outsiders um they're very clickish like

play09:15

that but at the same time I think that

play09:18

we're going to start seeing a little bit

play09:19

more jobs dislocation in 2025 um as the

play09:23

uh capabilities of these models expand

play09:25

but again there's a lot of skepticism um

play09:28

in the SE Suite of Enterprises and I

play09:30

don't just mean Tech Enterprises I mean

play09:32

you know Finance law like across the

play09:34

board manufacturing a lot of them are

play09:36

watching it and they're like oh yeah

play09:38

that's interesting we'll check on it in

play09:39

a couple years so that's that's why it's

play09:42

like you know sorry to throw a bucket of

play09:43

cold water on it but even if we get AGI

play09:45

tomorrow it's going to take a few years

play09:47

to get fully um integrated into um into

play09:51

the world and into the economy so that

play09:52

we can all you know live post- labor

play09:55

economic you know utopian Lifestyles um

play09:57

but I do suspect that uh 2025 will be

play10:01

the year that a lot of our benchmarks um

play10:03

are broken and then we'll have to have

play10:05

an entirely new set of benchmarks So

play10:07

based on all of these Trends 2026 is the

play10:09

year that I think that um that these

play10:11

models are going to be considered

play10:12

Enterprise ready um so there we're going

play10:15

to be on the other side of the trough of

play10:17

disillusionment um particularly with uh

play10:19

you know Claude six or you know uh sorry

play10:22

Claude 5 or GPT 6 or 5.5 or whatever the

play10:26

models are I think that that was two

play10:28

more Generations from now now are going

play10:30

to be when when all of the Enterprises

play10:32

on Fortune 500 all around the world um

play10:35

are going to be saying this is

play10:36

Enterprise ready we are ready to go and

play10:39

these are going to be what are what are

play10:40

going to be considered the first true

play10:41

general purpose models and so what I

play10:44

mean by general purpose models is

play10:46

basically kind of what Nvidia is working

play10:47

on which is like any like X tox or any

play10:50

to any modality um because then you just

play10:53

have a model that is a it is a ready to

play10:55

go off-the-shelf Droid brain you can put

play10:57

the same model in a car a CH y a digital

play11:00

agent those sorts of things and that's

play11:02

really what we're aiming for is those

play11:04

general purpose models and I don't mean

play11:05

just you know just text just a few tools

play11:08

just audio video I mean like anything

play11:11

geospatial data sensory data embodiment

play11:14

data everything to everything is what

play11:16

I'm really looking for for them to be

play11:17

truly Enterprise ready and general

play11:19

purpose and by then people might be

play11:21

saying yeah this basically qualifies is

play11:23

Agi honestly I wouldn't be surprised if

play11:25

we call those early AGI and then what

play11:28

we're talking about for AGI in 2027 2028

play11:32

actually constitutes ASI because like

play11:34

think of it this way if you have a

play11:36

general purpose model that you can put

play11:37

in a robot and that robot can perform at

play11:39

the level of like basically any person

play11:41

with 5 years of experience I think that

play11:43

would probably constitute AGI but then

play11:46

if you have you know two more

play11:47

Generations later you have a model that

play11:49

you can do anything with and it's

play11:51

basically like every PhD on the planet I

play11:53

would consider that super intelligence

play11:55

so you know your mileage may vary

play11:57

depending on definition but really from

play11:59

a from a commercial economic standpoint

play12:01

and and downstream from that from a uh

play12:04

from the impact that you will feel

play12:06

Enterprise ready general purpose models

play12:08

is really where it's going to be at and

play12:10

this is where um people are really going

play12:12

to start taking notice um around the

play12:14

world um now that's also two years down

play12:17

the road from now so that'll be midterm

play12:19

elections time so I wouldn't be

play12:20

surprised if if AI starts being talked

play12:22

about more in politics than it is

play12:24

already being talked about also by that

play12:27

time I wouldn't be surprised like if you

play12:29

know you many of us have domestic uh

play12:32

robots helping with cooking and cleaning

play12:34

by that point um they're probably going

play12:36

to be pretty expensive I know like the

play12:38

first gen are going to be like $80,000

play12:39

which it's like you could buy a Mercedes

play12:41

for that uh a nice Mercedes um or a nice

play12:44

truck like a really nice truck um but

play12:47

there's also going to be so much

play12:49

competition because the me the the the

play12:52

electromechanical aspect of these robots

play12:54

not particularly expensive and once you

play12:56

get to economies of scale the chassis is

play12:58

going to be $2,000 to maybe $5,000 for a

play13:02

good one um but then it's the software

play13:05

that you put in it it's the brain and

play13:06

the integration and and the testing to

play13:09

make it useful so I wouldn't be

play13:11

surprised if 2026 is probably the year

play13:13

that like I buy my first domestic

play13:14

assistant robot maybe before then I

play13:17

don't know I know that there are some

play13:18

other YouTubers out there that have

play13:19

bought like the the very first ones and

play13:22

they you know it's like cool you test it

play13:23

but it's not really useful it's not

play13:25

really a good product Market fit yet the

play13:28

2027 is expected to be the year of AGI

play13:31

so and again confidence uh intervals

play13:33

vary quite a bit so you know some people

play13:35

say like 2026 5% 2027 50 to 90% 2028 to

play13:41

2029 it's like seems like a foregone

play13:43

conclusion at least for some industry

play13:45

insiders that we will have AGI by then

play13:47

but again like I said it depends on your

play13:50

definition um I wouldn't be surprised if

play13:52

2027 2028 is when those general purpose

play13:55

Enterprise models are all PhD level now

play13:58

even when you have that it's still going

play14:00

to take time because there is so much

play14:02

inertia for for adoption um like

play14:06

governments will be paying attention

play14:07

militaries will be paying attention all

play14:09

corporations will be paying attention um

play14:12

and what I want to point out here is

play14:13

that once that excitement builds back up

play14:16

investors will be pushing for Rapid AI

play14:18

adoption meaning that all CEOs are

play14:21

basically going to be incentivized to go

play14:22

as fast as possible which is why I

play14:24

always say that acceleration is kind of

play14:26

the default policy the same thing is

play14:28

true on the Geo political stage um

play14:30

America is going as fast as possible uh

play14:32

China is going as fast as possible

play14:34

everyone is going as fast as possible to

play14:36

develop Ai and once you get to the point

play14:39

where AI is actually modifying the

play14:41

landscape the competitive landscape

play14:42

whether it's free market economics or

play14:44

military landscape everyone is going to

play14:46

be locked into that arms race now uh

play14:49

once you get to this point um there's

play14:52

like some of the some of the things that

play14:53

are going to slow down is going to be

play14:54

regulatory constraints um internal

play14:57

adoption policies safety res search um

play15:00

but you're also going to see a lot of

play15:01

creative creative disruption I would not

play15:03

be surprised if 2026 or maybe even 2027

play15:07

is when we see the first like

play15:08

featurelength Hollywood Blockbuster film

play15:11

created entirely by AI we've got a

play15:14

little bit of time before you know like

play15:16

we're already seeing people stitching

play15:18

together you know like commercials and

play15:19

stuff with AI but that's a human doing

play15:22

it and editing it what I mean is like

play15:24

end to end production you push a button

play15:26

and you get a you know you get a outp a

play15:29

4K Hollywood grade you know movie on the

play15:31

other side you know a couple hours or a

play15:33

couple days later however long it takes

play15:35

um it'll be a little while before models

play15:37

are that good where they can produce

play15:39

something that is not just you know some

play15:41

postmodern art housee garbage um that

play15:44

doesn't really make any sense because we

play15:45

can do that now with AI but what I mean

play15:48

is something that is going to like make

play15:50

a billion dollars at the box office um

play15:53

2027 I'm not going to put money on that

play15:55

but like I wouldn't be surprised if the

play15:57

first billion dollar uh box office

play16:00

Blockbuster happens in 2027 or

play16:02

thereabouts and that will kind of

play16:04

coincide with you know AGI or whatever

play16:07

um and we're going to see a lot more

play16:08

again we're going to see broader impacts

play16:11

we might start to see some layoffs but

play16:13

really I think that it's going to be a

play16:14

hot button political issue for the

play16:16

following year the next election cycle

play16:18

So speaking of the election cycle I

play16:20

think that 2028 is going to be a really

play16:22

spicy year um there's a lot that's going

play16:25

to be happening uh around that time so

play16:28

having watch watch a lot of uh military

play16:31

insiders um interviews with generals

play16:33

Admirals and other people from the

play16:34

intelligence community and just

play16:36

following closely on kind of the longer

play16:38

term uh trends that are happening in

play16:41

geopolitics I think that 2028 is

play16:43

basically going to be a hey we're

play16:45

gearing up for a new arms race we're

play16:47

gearing up for a new conflict with China

play16:50

um at the same time we're going to start

play16:51

to see those Mass layoffs because again

play16:54

once you get AGI you still need to you

play16:57

still need to invest in it you still

play16:58

need to deploy you still need to

play16:59

integrate into it so I think that we're

play17:01

going to be seeing like kind of having

play17:03

this onew punch of a new cold war or

play17:06

maybe even a hot War we'll see but new

play17:09

geopolitical uh uh Conflict at the same

play17:12

time that we're going to start seeing

play17:13

Mass layoffs due to uh the integration

play17:15

of AGI and Robotics um and so it's going

play17:18

to be a very very contentious uh

play17:20

election cycle and I wouldn't be

play17:22

surprised if we're talking about AI

play17:24

safety job protection Universal basic

play17:27

income and a whole bunch of other stuff

play17:29

and so it's basically going to be a

play17:30

fever pitch crisis year and I'm not

play17:33

basing that just off of my own stuff in

play17:35

fact before I was reading you know stuff

play17:38

like um the fourth turning and Ray Doo's

play17:41

work and talking to a few other people

play17:43

you know it's like well the it seems

play17:46

like everything is going to be coming to

play17:47

a head sooner rather than later and I

play17:49

think that 2028 might be kind of the

play17:51

inflection point another thing is that

play17:53

there's a lot of commentators that

play17:55

suspect that conflict with China is

play17:56

going to happen between sometime between

play17:58

2026 and 2032 and that election year you

play18:02

know the the Advent of AGI robotics you

play18:05

know Quantum Computing and fusion um

play18:08

kind of on the horizon that really could

play18:10

be the inflection point and one other

play18:12

thing is that the clock is ticking for

play18:13

China because they're facing demographic

play18:16

collapse um basically the the their

play18:19

their demographic pyramid is inverted

play18:20

right now because they're having so few

play18:22

births and it actually looks like if you

play18:24

watch the news coming out of China very

play18:26

closely it looks like they might might

play18:28

have been lying about their population

play18:29

numbers for several years so China's

play18:32

population might already be shrinking in

play18:34

the order of tens of millions to even

play18:36

prot potentially hundreds of millions

play18:38

per year again it depends on who you

play18:40

listen to but it seems like a foregone

play18:42

conclusion that China's population is

play18:44

shrinking and that basically means that

play18:47

the clock is ticking for them any big

play18:49

moves that they want to do they need to

play18:50

do sooner rather than later while they

play18:52

still have the manpower to do it um

play18:54

which means that that could move up any

play18:56

timeline for any potential conflict with

play18:58

China and it doesn't necessarily have to

play19:00

escalate to a hot conflict and again all

play19:02

that's going to be going on on the

play19:04

geopolitical stage and uh and in the

play19:07

meantime we're going to be having

play19:08

domestic issues potentially with layoffs

play19:11

and you know the economy is going to be

play19:13

on fire and yet wages are going to be

play19:15

stagnating total employment might be

play19:17

going down by then unemployment might be

play19:19

going up um and so there's going to be a

play19:21

lot of anger um around this time I'm not

play19:24

going to say that it's going to be like

play19:25

you know the Next Great Depression or

play19:26

anything like that because GE political

play19:29

conflict has a tendency to um do good

play19:32

things for jobs in the economy which you

play19:34

know I wish that weren't true but you

play19:35

know the military in Jal complex exists

play19:38

you know don't shoot me I'm just the

play19:40

messenger now 2029 I think is going to

play19:42

be um really when most people re like

play19:46

kind of his history will will record

play19:48

2023 is the year that it started 2024

play19:51

the year that it accelerated but I think

play19:53

that 2029 is going to be the year that

play19:55

everyone is like yes this is the

play19:56

beginning of the New Renaissance because

play19:59

this is when all the gigantic new data

play20:00

centers are going to come online this is

play20:02

when um the commercial nuclear fusion

play20:04

reactors are expected to see first light

play20:07

this is when Quantum Computing is going

play20:08

to become mainstream and then all the

play20:10

downstream effects of that namely

play20:11

Material Science and genetic engineering

play20:14

will basically you'll start to see that

play20:16

flywheel accelerate and you'll start to

play20:18

see real true compounding returns um

play20:20

from all of these Technologies mixing

play20:22

together and you know it might be a

play20:24

foregone conclusion by then that like we

play20:26

have ASI or whatever um robots are going

play20:29

to be much more commonplace we're going

play20:32

to really start to see a major shift in

play20:34

medical breakthroughs longevity escape

play20:36

velocity is something that people are

play20:37

going to be talking about um you're

play20:39

going to see more cybernetics those

play20:41

sorts of things but this is basically

play20:44

like the year that it all begins or the

play20:45

year that it really changes and from a

play20:48

from an individual perspective I suspect

play20:50

that this is when we're going to start

play20:52

to see a new level of optimism um kind

play20:55

of akin to the post-war boom of the

play20:57

1950s hopefully it doesn't it's not

play20:59

delayed like cuz a con a hot conflict

play21:02

with China could be devastating to not

play21:04

just the two Nations but to the whole

play21:06

world so I hope that it doesn't escalate

play21:08

to a hot conflict and it's entirely

play21:10

possible that it won't um the way that

play21:13

America and China are communicating

play21:15

makes me think that it's just going to

play21:16

be another cold war um which okay that's

play21:19

not the best thing in the world um but

play21:21

that's certainly far preferable to a hot

play21:23

conflict um but then at the same time

play21:26

all of those you know like basic think

play21:29

of the optimism of the 60s and 70s

play21:31

during the space age right we're going

play21:32

to have that level of optimism again and

play21:34

again it's not just me predicting that

play21:36

uh Ray doio fourth uh the fourth turning

play21:39

a lot of these theories think that we're

play21:40

that we're at or near an inflection

play21:42

point and that after the inflection

play21:44

point it's going to be much much better

play21:46

in the future now obviously I'm not one

play21:48

to live in the future but you know we

play21:50

all need something to hope for and we

play21:51

need some optimism um so I think 2029 is

play21:55

is really kind of targeting that when uh

play21:57

Enterprise level adoption is going to be

play22:00

accelerating geopolitics is going to be

play22:02

changing and then that will lead into

play22:04

2030 which is kind of the year that it's

play22:06

like okay this is going to be the new

play22:08

normal so in 2030 I suspect we will have

play22:12

coined a new term we might call it the

play22:14

intelligence age because you know I was

play22:15

just referring to like the Space Age so

play22:17

you know the 60s 7s um I guess early in

play22:21

the or late late in the 50s 60s and 70s

play22:24

that was like the space age and then the

play22:26

80s and 90s that was you know the

play22:28

Information Age um and now we've had the

play22:31

Grim Dark Age of the 2000s and 2010s and

play22:34

early 2020s I think that we're going to

play22:36

be entering into entering fully into the

play22:38

new paradigm of the intelligence age or

play22:40

the AI age um by 2030 this is where

play22:44

we're going to be basically kind of

play22:45

settling into the new normal of the

play22:47

fourth Industrial Revolution um which is

play22:49

why I frequently kind of compare 2030 to

play22:51

like 2050 or 1950 because these Cycles

play22:55

generally come in about 60 to 80 year

play22:56

Cycles sometimes a little bit more

play22:58

sometimes a little bit less so we are

play23:00

basically overdue or almost overdue for

play23:03

a new like New Renaissance or new um

play23:07

Golden Era and so I'm hoping that by

play23:09

2030 we are entering into that new

play23:11

golden era and I've been I've been

play23:12

pretty consistent on this messaging for

play23:14

a while that the next 5 years are going

play23:16

to be some of the hardest most painful

play23:19

years because old paradigms are going to

play23:21

stop working and we're going to start to

play23:23

need new paradigms namely post labor

play23:25

economics so when the first time that

play23:28

bloom ber and Forbes and all those other

play23:29

ones um talk about post- labor economics

play23:32

I will be super excited um but it seems

play23:35

basically inevitable uh that the that

play23:37

economics as we understand it today will

play23:39

just not work I do think that money is

play23:41

going to stick around and I think

play23:42

capitalism is going to stick around but

play23:44

I think that neoliberalism is on its

play23:46

last legs um and that's going to go the

play23:48

way of the dinosaurs anyways that's

play23:50

personal speculation um longevity escape

play23:53

velocity Ray Kurtz predicts that

play23:55

longevity escape velocity will come by

play23:57

2030 and as many of you in the audience

play23:59

say don't ever bet against Ray kurts now

play24:02

I will say having read some of his older

play24:03

books some of his predictions were dead

play24:05

wrong but some of his predictions were

play24:07

were pretty much spoton um so he his his

play24:10

batting average is like you know 6.7 um

play24:13

which is certainly good given the time

play24:16

Horizons that he's been working on um

play24:18

but this is one where I I tend to agree

play24:20

that you know longevity escape velocity

play24:22

all those compounding returns from AI

play24:24

Quantum Computing and all the downstream

play24:26

effects of that I I think that we could

play24:29

achieve longevity scape velocity by 2030

play24:32

um the socioeconomic shifts I think that

play24:35

we'll we'll probably have something like

play24:37

Ubi by then um but we'll also realize

play24:39

that Ubi alone is not enough we'll need

play24:41

an entirely new approach to um finding

play24:44

meaning uh giving people stuff to do

play24:46

that sort of thing um I also hope that

play24:51

this new era of abundance will reshape

play24:53

the geopolitical conversation but it

play24:55

will also be a little bit early and what

play24:57

I mean by early is that even if we all

play25:00

have AGI and Hyper abundance it's going

play25:03

to take several Generations if not

play25:05

longer I mean it could take several

play25:07

centuries to really kind of heal the

play25:09

planet and have everyone come together

play25:11

um because there's just so much distrust

play25:14

and so much inertia so much geopolitical

play25:16

inertia out there um I certainly hope

play25:19

that we don't have any more Wars once we

play25:21

have nuclear fusion in AGI because you

play25:24

know the weapons that that are capable

play25:27

of being produced then who was it I I

play25:29

think someone said was it James Cameron

play25:31

I don't remember someone said I don't

play25:32

know what weapons World War III will be

play25:34

fought with but World War I will be

play25:36

fought with sticks and stones that's

play25:38

really kind of the Paradigm that we're

play25:39

looking at if I hope that we don't have

play25:41

a World War III and if we don't then I

play25:44

think we're going to be um all all I'm

play25:47

not going to say happily ever after

play25:48

because we're humans we're good at

play25:49

making a mess of things but I do think

play25:51

that um I do think that the hardest

play25:53

point is in the uh years just ahead and

play25:56

then once we get past that I think it'll

play25:58

be a little bit easier for a while so

play26:01

this is these are how I actually feel

play26:03

like things are going to progress over

play26:04

the next 6 to seven years let me know

play26:06

what you think in the comments um yeah

play26:08

no this is I I hope that this kind of

play26:10

gets everyone on the same page again

play26:12

because yes timelines are a little bit

play26:13

longer than I'd hoped but I think that

play26:15

also like what's that rule where people

play26:17

say it's like people always overestimate

play26:19

the short-term impact but underestimate

play26:21

the long-term impact I'm hoping that my

play26:24

predictions are getting a little bit

play26:25

more grounded in reality and a little

play26:27

bit more accurate to how things are

play26:29

actually going to play out so thanks for

play26:31

watching cheers have a good one

Rate This

5.0 / 5 (0 votes)

関連タグ
Artificial IntelligenceFuture PredictionsEconomic ShiftTechnological AdvancementRoboticsAI DisillusionmentEnterprise AdoptionGlobal PoliticsPost-Labor EconomicsLongevity VelocityNew Renaissance
英語で要約が必要ですか?