What this covers
Ray Kurzweil, computer pioneer and futurist, presents a hopeful view of what AI can achieve in the future in a wide array of fields, from language translation to medicine, energy, education, transportation, and farming. He argues that by 2029, AI will achieve human level intelligence and that “most diligent people” will reach “longevity escape velocity,” (the ability to add time to our lives through technological advancements at a higher rate than our longevity is being used up). Learn more about COSM2024 at https://cosm.tech. ------------------------------------------------------------------------------------------------------ The mission of the Walter Bradley Center for Natural and Artificial Intelligence at Discovery Institute is to explore the benefits as well as the challenges raised by artificial intelligence (AI) in light of the enduring truth of human exceptionalism. People know at a fundamental level that they are not machines. But faulty thinking can cause people to assent to views that in their heart of hearts they know to be untrue. The Bradley Center seeks to help individuals—and our society at large—to realize that we are not machines while at the same time helping to put machines (especially computers and AI) in proper perspective. For more about the Bradley Center visit https://centerforintelligence.org/. Be sure to subscribe to the Center for Natural and Artificial Intelligence on Youtube: https://www.youtube.com/channel/UCjmYJ4vVctTLr5Xn-K1EsYA
Follow Walter Bradley Center for Natural and Artificial Intelligence on Facebook and Twitter: https://twitter.com/cnaintelligence, @cnaintelligence Facebook: https://www.facebook.com/BradleyCenterDI/
Visit other Youtube channels connected to the Discovery Institute:
Discovery Science News (Center for Science & Culture): https://www.youtube.com/DiscoveryScienceNews
Dr. Stephen C. Meyer: https://www.youtube.com/user/DrStephenMeyer
Source description (no synthesized summary yet).
Ray Kurzweil argues that exponential growth in computational power will enable AI to achieve human-level intelligence by 2029, solve major diseases through protein folding and biomedical simulation, and allow humans to reach 'longevity escape velocity' by extending lifespan faster than time passes, fundamentally transforming human existence through merger with technology.
- Computational power has doubled exponentially for 80 years at a predictable rate, reaching a tipping point that enables large language models and AI applications across medicine, manufacturing, and energy.
- AI will accelerate medical breakthroughs by 1,000x through protein folding prediction (AlphaFold 2) and biomedical simulation, eliminating aging and most diseases within 10 years.
- By 2029, scientific progress will extend remaining lifespan by one full year for every year lived, creating indefinite lifespan extension without requiring biological implants, merely cloud-connected enhancement.
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Computational power has followed an exponential growth curve for 80 years (from 1939 to 2023), with each level on the chart representing 10x improvement; Kurzweil tracked this trend for 40 years and found it represents a smooth, predictable exponential curve uninterrupted by major world events like World War II or economic depressions.
“a straight line represents exponential growth but look at how predictable this is I started analyzing this 40 years ago it's a smooth Trend with no interruptions despite major world events like World War II or economic depressions you don't see any example of that on this graph”
Large language models have hallucinations (generate false information) because they learn from human-generated data and humans are not entirely accurate; this is acknowledged as a limitation but also reflects the human condition—people transmit and propagate inaccuracies.
“the reason that we have what's called hallucinations in large language models is because it learns from people and people are not accurate that's quite an admission Ray that that if people go Ary it in in intoxicates the large language models and we're going to blame the failures of the singularity on the people who have created it and continue to maintain it and contribute to it”
Machines have historically extended human abilities—buildings couldn't be built without tools, creating buildings requires technology—showing that humans have always merged with technology to advance ideas; creating AI to extend brains is the same process as using machines to build buildings.
“this is exactly what we've done in the past we've created machines to help us do things like create buildings I mean you're in a building could you created this building yourselves uh obviously we use technology to advance our own ideas these machines enhanced our abilities we've learned to trust them now we're creating AI to extend our brains in the same way”
Moore's Law is commonly cited for computational growth, but it is actually narrower and less accurate than the true exponential growth trend: Moore's Law applies only to integrated circuits (and Intel has only ~10 of the 80 best computers by computational power per dollar); the broader exponential trend predates integrated circuits by decades and encompasses multiple technological paradigm shifts.
“people have called this M's law that's really not correct because Mo's laws has to do with integrated circuits it started actually decades before intel was even formed of the 80 best computers in terms of computations for second per dollar only 10 of them have anything to do with int Intel”
The mRNA COVID-19 vaccine was created in just two days by testing several billion different possible inhibitor compounds using AI, compared to the traditional method of testing one compound per year and requiring many years (often 10+) to develop a medication; subsequent human testing required 10 months.
“many of you have probably taken the Mna covid vaccine uh that was actually created in two days uh we actually wrote down all the different things that might uh inhibit Co vac vacine it was actually several billion different uh possible things we actually test the way it used to work is we would test one thing would take a year then we might modify it take another few years and if you're lucky over 10 years maybe You' get a medication they tried all several billion and that took two days”
Most people do not accept or understand exponential growth despite its apparent visibility in empirical data, and economic models continue to use linear assumptions about the future, contributing to widespread misunderstanding of technological trajectories.
“first of all most people actually don't accept the exponential growth it seems quite obvious you look at the day data it's very very apparent I mean that graph shows complete exponential growth of computers over 80 years and it's true of everything else uh but people actually don't accept that economic models use linear ideas of the future”
Throughout history, humans have not had the means to share insights between individuals—even brilliant individuals in the past wouldn't be known to others due to lack of communication technology; the ability to fill our brain with true knowledge required technology like written language to propagate understanding across populations.
“if you go back a thousand years before people had written language um people didn't really know that much because they weren't able to share insights so even though a human being would have an Insight another human being would wouldn't know about it because we didn't have the means of communicating it so just filling up our own brain with things that are true requires technology”
In 1800, approximately 80% of jobs involved livestock and farming; today only about 2% of the workforce is involved in farming, yet the world produces far more food than in 1800, demonstrating that automation and innovation eliminate old job categories but create new ones at faster rates.
“in 1800 I could go through all the jobs 80% of which had to do with with uh livestock and farming uh and I would say all these jobs are going away because we now have two% dealing with farming and yet we produce much more food um so we're actually and so people say where's our jobs going to come from”
AlphaFold 2's protein structure predictions unlock the ability to synthetically produce proteins with desired functions, such as proteins that stimulate the immune system to fight cancer, which was not possible even five years ago (pre-2018), enabling new therapeutic approaches.
“this marked a profound step forward in medicine because it unlocked information that need to synthetically produce proteins with the desired function uh for example proteins that will stimulate our immune system to fight cancer and we're actually seeing that now and this was not possible even five years ago”
A questioner proposes an alternative to the Turing Test: a large language model cannot and will not rival human intelligence unless it can unilaterally decide to ignore its node weights while consistently providing coherent responses, meaning true intelligence requires the ability to violate its own programmed constraints.
“I would like to he your thoughts on an alternative to the touring test my assessment is that AI cannot and will not rival human intelligence unless and until a large language model can unilaterally decide to ignore its node weights while consistently providing cogent responses”
Creativity in artificial systems always comes as a surprise because fundamentally surprising information is the definition of creative output; large language models knowing everything humans know does not capture everything, because humans don't know most things we would like to know, and scientific progress will be accelerated using LLM methods.
“creativity always comes as a surprise to us right create information is fundamentally surprising bits as you know and uh and actually if if a lar language model knows everything that people know that's not everything because we don't know everything in fact most of the things we'd like to know we don't know”
Kurzweil responds that one person knows very little—even Einstein knew physics but not philosophy and psychology—and that large language models can combine and synthesize the knowledge of all humans across all domains, which is what is truly valuable rather than individual knowledge.
“if if you take one person uh that one person doesn't know everything in fact they know very little even if you take somebody like Einstein you knew a lot of things about like kwell uh he knew certain things about physics but he didn't know things in philosophy and psychology uh what we really want to master is everything that all human beings have done that's what large language models can do you can can ask it anything and we will give you a very intelligent answer uh one person is fairly limited”
In the past 13 years (2010-2023), the amount of computation devoted to training the best AI models has doubled on average more than twice per year, representing a 10-billion-fold increase since 2010; this acceleration explains why large language models emerged only in the past year despite AI research spanning decades.
“in the past 13 years the amount of computation devoted to training the best AI models has on average doubled more than twice a year that's a 10 billion millionfold increase just since 2010 so I mean a year ago it was not uh strong enough to do large language model now a year later uh that's taking over AI”
An immunotherapy drug called Dostarlimab, tested in a small clinical trial of 14 colorectal cancer patients at Memorial Sloan Kettering Cancer Center, produced unprecedented results: after six months of treatment, all 14 patients, including those with stage 4 cancer, had their tumors completely disappear and the tumors have not returned—the first time complete remission has been achieved in a clinical trial.
“we're already making profound progress in fighting cancer uh one exciting advancement for example is an immunotherapy drug called doab it was tested in a small clinical trial of 14 colar rectal patients by Memorial slone kering cancer center after six months of treatment every single patient on the study received the drug including people with stage four cancer had all their tumors disappear and they have not come back this is the first time that's ever happened usually you make 20% gain or 30% gain every single one went away”
A poll surveyed 24,000 people across 23 countries on whether extreme poverty has improved or worsened over the past 20 years; 70% thought it had gotten worse, only 12% thought it had gotten a bit better, but reality shows poverty has actually decreased by 50% over the past 20 years, a result predicted by only 1% of respondents.
“this chart shows the result of a poll it was um 24,000 people in 23 countries they were asked whether extreme poverty has gotten better or worse over the past 20 years 70% thought it had gotten worse only 12% thought it had gotten a little bit better the reality is that poverty has actually decreased by 50% over the last 20 years and that was a prediction of only 1% of the people who responded”
Work weeks have declined over time as a consequence of increased automation, representing another dimension of human progress; despite automation reducing labor hours, more people are employed overall than in the past.
“our income in constant dollars has gone up all our work week has gone down um it's shot up slightly after the Great Depression but it's continued to go down since then this is one consequence of increased automation but we actually have more people working than we than we've had in the past”
For 50 years, scientists were aware of the protein folding problem: knowing that a one-dimensional amino acid sequence folds into a three-dimensional protein structure, but a small number of research groups were able to predict these structures with only ~20% accuracy before AlphaFold 2 was created.
“for 50 years scientists were aware of this uh we actually produce a one-dimensional sequence of amino acids and finally at a certain point that the linear sequence of amino acids folds up into a three-dimensional protein it's called the protein folding problem and people have actually studied this a small number of research es were able to do this they weren't very successful getting maybe 20% of them correct”
Life expectancy improvements largely resulted from developing methods to avoid or kill external pathogens (bacteria and viruses) that cause disease from outside the body, addressing the major causes of death (like cholera and dysentery) that were particularly lethal in infancy and youth.
“this was largely achieved by developing ways to avoid or kill external pathogens bacteria and viruses that bring disease from outside our bodies um and I could go on about that this chart shows average personal income in constant dollars this is not to do with inflation”
In the energy sector, if humans captured just one part in 10,000 of the solar radiation falling on Earth, it would meet all of humanity's energy needs; currently, humans are creating more energy through solar and wind than through oil and gas annually, and this is following an exponential curve that will enable meeting all energy needs through renewable sources by the 2030s.
“for example in in energy we're actually if we uh were to take one part in 10,000 of the that falls on the earth we'd meet all of our energy needs and we're actually now uh we're creating more energy through solar and wind than we are through oil and gas in this past year and it's actually an exponential curve by the 2030s will be able to meet all of energy needs that way”
Human-operated cars cost the United States 40,000 lives per year and millions of lives globally, but with self-driving cars this number will decline by about 99%.
“we're also doing away with many accidents for example human operated cars cost us today just in the United States 40,000 lives a year in this country it's millions of lives across the world world but with self-driving cars this number will decline by about 99%”
A female questioner asserts that the true singularity is not merging with machines but harnessing unharnessed technology within human brains; her research team of girls has discovered a new language to use brain parts that remain unharnessed, focused on becoming more alive rather than living longer, using a non-time/space-based language of consciousness.
“I believe the true Singularity is not merging with the machine that it is harnessing unharnessed technology in our own brain our own brain is the most spectacular technology made so we discovered as girls a new language to use parts of our brain that are unharnessed I'm interested this language is about becoming more alive not living longer it is not a language of time space”
Damis Hassabis, who headed the AlphaFold 2 work at Google DeepMind, has launched a new lab using AI to process and model vast genetic data for pinpointing medical solutions in 1/1000th the traditional time.
“damis hbus who headed up this work at UH Google Deep Mind is launched a new lab that's going to perfect this um to using AI they're processing and modeling vast amounts of genetic data to pinpoint Medical Solutions in 1,000th the time it it used to take traditional methods”
Kurzweil responds that the same brain mechanisms (neural connections) that enable human creation can be emulated in machines; large language models make connections between facts, which is identical to what happens in human brains, suggesting machines can achieve abstraction.
“all happens in our brain our brain has different connections large language models have same type of connections they can do the same thing we can do they can do it much faster so I gave the example where uh we uh tried every single possible way in which we could inhibit a co vacine uh and it did it in two days”
An open-source market will emerge for printed clothing, housing, and food created with AI and vertical agriculture, democratizing access to basic material needs and creating economic leveling by eliminating scarcity in essential goods.
“there'll be an open- source market for printed clothing housing food create with the help of AI vertical agriculture uh this will be a great economic leveler”
The computational power of the Google Cloud TPU V4 achieves 50 billion calculations per second per dollar in constant dollars, compared to 0.007 calculations per second per dollar in 1939—an advance that represents the progression from telephone relays to vacuum tubes to transistors to integrated circuits to cloud-based AI systems.
“the chu goes up on the upper right hand corner to the Google Cloud TPU V4 which for the same amount of money set of 0.007 calculations per second in 1939 it's now 50 billion calculations per second per dollar so that's qu quite uh an advance all with constant dollars in the past 80 years”
The Turing Test as originally written by Alan Turing in 1950 was poorly described—only a few sentences—and its meaning is ambiguous with different interpretations; GPT-4 has already been assessed as passing the Turing Test, and over the next few years multiple systems will be reported as passing before formal assessment settles the question.
“I've actually felt when Turing wrote this the Turing test in his paper in 1950 he didn't describe it very well it's only a few sentences and there's different ways of interpreting it yeah uh and so I felt we'd actually have a few years where people will be reporting that a computer has passed the TR test uh and it will take a few years till we actually assess it accurately”
AI will achieve human-level intelligence by 2029, a prediction Kurzweil made 24 years ago in 'The Age of Spiritual Machines' (1999); when AI achieves human-level capability, it will simultaneously surpass humans in every way, and humans will merge with AI rather than be left behind.
“AI is exactly where it should be right now to achieve human level intelligence by 2029 so that's something I predicted 24 years ago in my book the age of spiritual machines so once AI reaches human level capability it will at the same time soar past us in every way”
George Gilder counters Kurzweil's 2029 human-level AI prediction by arguing that despite exponential computational growth (which Gilder acknowledges), the Singularity has already happened computationally but has 'scarcely impinged' on human nature itself—computation and human consciousness are separate paths.
“certainly by every computational measure your law of exponential returns accelerating returns has uh prevailed and our computer technology does Ed Us in all these computational applications millions of applications of PL proliferated and so I I believe that in fact uh The Singularity has happened and uh yet it's scarcely impinged on us as human beings and our uh intrinsic Natures at all it's a separate path”
Gilder objects that Kurzweil's vision implies a singular 'Singularity of Truth' in the cloud, as if intelligence becomes one vast collective unified entity; but human minds are connectomes (estimated at 2 zettabytes to map by MIT researcher Sebastian Song) containing the contents of the entire global internet, suggesting intelligence is distributed and independent across billions of minds.
“but is it unitary Ray your image implies a sort of Singularity of Truth in the cloud as if uh intellig ience is some vast Collective Gomer uh that uh has attained some s sort of perfection human Minds which uh are connectomes of as Sebastian song of uh MIT is estimated the connectome of the human brain is uh takes two zettabytes to map which is about the contents of the entire Global Internet”
George Gilder raises a distinction between 'intelligence' (processing information) and 'abstraction' (conceiving, designing, theorizing new things); machines can be made to process information faster and better than humans, but the question remains whether machines can invent, conceptualize, or abstract beyond human capabilities.
“I know um um start with information that exists and um what humans can do goes far beyond that we we can uh conceive design uh theorize uh things of this sort which I call abstraction as opposed to intelligence which is processing information and I don't and we understand the mechanisms for doing that we understand Compu um uh logic and and ways you know we understand uh uh neural networks and so on and we can emulate them and we we make machines that can do it faster and better than we can do it ourselves but machines how how do you make a machine that can do invent things uh conceptualize and so on”
In the 2030s, nano-robots will non-invasively enter the brain through capillaries (without surgery) and provide wireless communication between the neocortex (top layer of brain) and cloud-based computing, similar to how modern smartphones extend cognition through cloud connection.
“in the 2030s Nano robots will go into our brains non-invasively without surgery through the capillaries provide wireless communication between the top layer of our brain the neocortex and in the cloud uh just the way your uh smartphone does that today with the cloud”
AI will spread through every industry beyond medicine, producing breakthroughs in manufacturing, energy, farming, and education; education is described as 'very old-fashioned' and requires significant acceleration; once reaching 2029, the world will change very rapidly.
“AI spreading through every industry not just medicine we're going to see breakthroughs in lots of different fields including um manufacturing energy farming education education is very old-fashioned we really need to speed up with that once we get to 20 29 the world is going to change very very rapidly”
Smartphone adoption demonstrates the rapid adoption of brain-cloud extension: 6-7 years ago only ~33% of people raised hands when asked if they had a smartphone; recently, everyone raised their hands and no one raised their hand when asked who doesn't have a phone, showing smartphones are already extending human cognition.
“today and we actually use this six or seven years ago I asked somebody how many people have their smartphone and I'd say maybe a third of the people rais their hand I did this recently everybody raised their hand I said who doesn't have their phone nobody raised their hand so this is already extending who we are even though it's separate”
Large language models are not limited to language tasks despite their name; they are better understood as 'large advent models' applicable to medicine, manufacturing, energy, farming, and essentially every domain humans care about, not just linguistic processing.
“uh they aren't this limited to language they're actually large Advent models they can be applied to anything medicine I'll talk more about that because that's very exciting manufacturing energy farming basically everything we care about”
Historically, humans have had to compete to meet physical survival needs; as abundance becomes available through technology, the primary human struggle will shift from meeting physical needs to seeking purpose and meaning, making existential and psychological fulfillment the central challenge.
“historically humans have had to compete eat to meet the physical needs of life but as we enter an era of abundance when material Necessities are available to everyone our main struggle will be for purpose and meaning”
The term 'artificial intelligence' is misleading because it implies the intelligence from AI is not real, but AI-generated intelligence is actually real and based on human intelligence; large language models now demonstrate capabilities (like writing intelligently on any topic) that rival and exceed human abilities.
“I actually don't like that name because it implies that the intelligence we get from AI is not real but it is real it's based on on human intelligence and we're actually mastering human intelligence large language models are already better than what we can do I mean who can write something intelligent about anything you ask it”
One of the most profound near-term implications of exponential technological growth is vastly accelerated progress in creating new medications for diseases, which Kurzweil considers the most exciting opportunity for AI application.
“one of the most profound near-term implications of the exponential growth of technology is a vast acceleration of progress in creating new medications for diseases that's actually I think the most exciting opportunity”
Many people express pessimism about the future and are reluctant to have children because they believe things are worse now than ever before, but this perception is inaccurate; in fact, conditions are much better than they used to be, though not everything is positive.
“you know I hear people say they don't want to bring kids into the world because they think things are worse now than ever before that's really not accurate not everything is good but it's much better than it used to be people think that things are getting worse when they're actually getting better”
Business models must be rethought and entirely new models created that don't currently exist; this requires courage to let go of old linear assumptions about time and progress; future jobs won't resemble current jobs, requiring adaptation similar to how jobs 15 years ago (social media influencer) didn't exist 15 years prior.
“we need to rethink old business models and create new ones that don't exist yet it's going to take courage to let go of old linear assumptions people still think in linear they think something took 30 years in the past can take 30 years in the future and that's just not correct future jobs are not going to look like current jobs just think back 15 years ago who know that you could make money being a social media influencer and lots of other types of jobs that didn't exist 15 years ago”
Kurzweil clarifies he was not predicting 'immortality by 2029' but rather that by 2029, most diligent people will reach 'longevity escape velocity'—the point where they are adding more time to their remaining lifespan than is passing, thus beginning indefinite lifespan extension.
“you may have read in the press that I was predicting humans will achieve immortality by 2029 that's not exactly what I said as a result of protein simulation and simulating uh biomedical interventions by around 2029 most diligent people I think everybody here is a diligent person will reach what I call Longevity escape velocity the point where we're adding more time to our remaining lives than is going by”
As AI masters all human skills, individual human values will shift from evaluating people based on specific skill sets to valuing entrepreneurship, creativity, adaptability, and the ability to find and pursue personal passions and interests.
“artificial intelligence Masters all human skills values will shift from Individual skill sets to things like entrepreneurship creativity adaptability so to thrive in the future we'll need to explore our interests find passions that we care about”
Currently, as humans live through a year, they use up one year of longevity; however, scientific progress provides approximately 3-4 months of lifespan gain annually, resulting in a net loss of approximately 8-9 months of longevity per year lived.
“right now as you live through a year you use up a year of your longevity however science is also progressing so today you get back about 3 to four months from breakthroughs in scientific progress so you only really lose about eight or nine months of longevity in a year”
If AI is built to mirror human values and nature, humans will trust it as we trust other people (though noting that trusting other people 'isn't always a great idea'), and AI systems will have extensive protections to ensure truthfulness and respect for human values, similar to how humans raise children with knowledge, beliefs, and values.
“if we build it to mirror ourselves uh we will trust it and it will trust us and it will be part of Who We Are it'll be an extension of ourselves it will hold our values so trusting AI will be just like trusting other people which isn't always a great idea but you I think it will have a lot lot of protections I know people who are creating large language models they put more effort into making sure that what it says is true avoiding possible things that violate human uh uh skills you could compare to raising kids”
Search engines are generally considered quite accurate based on polling data showing people rely on them; search engines will be used to verify LLM outputs and ensure accuracy; this approach will improve LLM reliability, though people disagree about what constitutes 'accurate' information.
“what do you think about search engines uh they're generally considered quite accurate as we do polls people actually rely on search engines uh and we're actually now testing what a large language model will do based on search engines yeah so it'll have the accuracy of search engines which are actually pretty good um so large language models will become more accurate and they they already are becoming that way”
Literacy rates have risen globally from the 15th century onward; average years of education worldwide have increased since 1870; showing that educational progress is a key metric of human development.
“uh these are literacy rates by country going back to the 15th century um here are the average number of years of education around the world since 1870”
Nearly all functions of the body are carried out by proteins, which serve as the building blocks of life; the unique three-dimensional shapes of proteins dictate the success or failure of all medicine, making protein structure prediction the fundamental bottleneck in drug development.
“nearly all functions of the body are carried out by proteins uh they are the building blocks of life uh the unique individual shapes dictate the success or failure of all medicine”
Ray Kurzweil has been working on artificial intelligence for 61 years, longer than anyone else, having gotten involved in 1962, only six years before artificial intelligence was formally named as a field in 1968.
“I've applied this philosophy to artificial intelligence for 61 years uh that's actually longer than anybody else uh I got involved in 1962 only six years before that artificial intelligence was actually named”
The first programmable computer, the Zuse Z3, was created in 1939 by a German engineer (Zuse) who was not a Nazi sympathizer; it performed 0.07 calculations per second per dollar, and despite being shown to Hitler and the Nazi party, they saw no military value in computation—a catastrophic miscalculation.
“the first point here 1939 it was actually the first programmable computer called the zeusa 2 created in 1939 it performed 07 calculations per second per dollar so Zeus was a German he apparently was not a fan of Hitler but the computer was shown to Hitler and the Nazi party saw no military value to computation a very big mistake for them”
The Colossus computer, created by Alan Turing and colleagues with the backing of Winston Churchill, was instrumental in decoding Nazi messages during World War II; England had inferior air power compared to Nazi Germany but used the Colossus to win the Battle of Britain and provide the launching pad for the D-Day invasion, essentially saving England and Europe.
“the third computer on this chart is the Colossus created by Alan Turing and his colleagues so Winston Churchill immediately saw the value of this he got very much behind touring actually protected him and the Colossus computer uh he felt this computer would be key to winning World War II which it was the British got totally behind it and they use it to completely decode Nazi messages so everything that Hitler wrote or read was also read by Church”
Life expectancy has increased dramatically: 1,000 years ago it was in the 20s; by 1800 it was 35; by the mid-19th century in the UK and US it reached the 40s; and today it is approximately 80 in much of the developed world—nearly tripling in 1,000 years and doubling in the past 200 years.
“thousand years ago European life expectancy at Birth was just in the 20s actually thousand years ago uh life expectancy was 20 in 1800 it's 35 uh so many people died in infancy or youth from diseases like C and dissenter which are now easily preventable by the middle of the 19th century life expectancy in the UK and us had increased to the 40s uh now as a uh last year it is written to about 80 in much of the developed worlds so we we've nearly tripled life expectancy in the past thousand years we've doubled it in the past 200 years”
Average personal income in constant dollars (inflation-adjusted) has increased substantially since 1929, with income rising overall despite some fluctuations after the Great Depression; this growth reflects increased productivity and automation, not inflation.
“this chart shows average personal income in constant dollars this is not to do with inflation uh since 1929 uh if you go back 100 years certainly 200 years life was extremely difficult and there was no government programs that could support that uh our income in constant dollars has gone up”
Ray Kurzweil and George Gilder both received awards for entrepreneurial Excellence at the White House during the Reagan Administration, and have been close friends ever since, occasionally contending over various issues.
“I've known Ray I can't even think how long but it goes back to the Reagan Administration when I got uh we both that simultaneously received um awards for entrepreneurial Excellence at the White House”
Ray Kurzweil invented the first CCD flatbed scanner, the first omnifont optical character recognition machine, the first print-to-speech reading machine for the blind, the first text-to-speech synthesizer, and the Kurzweil music synthesizer (capable of recreating the effects of a grand piano so accurately it was tested against a real grand piano).
“he's uh of course you all know he's the world leading inv inventor uh what he invented the first CCD flatbed scanner did you know that of the first omn font optical character recognition machine uh the first print to speech reading machine for the blind this is really these are all the precursors of AI that uh Reay um uh created uh the first text to speech synthesizer and then he's famous of course for the KW music synthesizer which Stevie Wonder uh popularized”
Ray Kurzweil is a highly accomplished writer, having written bestselling books including 'The Singularity is Near' and 'How to Create a Mind' (both New York Times bestsellers), and also wrote a novel called 'Danielle' which won young adult reading awards.
“and he wrote books he is this is what really impresses me I'm a writer and Rey is just a terrific writer on top of all these Technical and mathematical and capabilities he is an amazing inventive original writer and the singularity is near and how to create a mind we terrific uh New York Times bestsellers but he also wrote Danielle you might want to read Danielle it won young adult reading Awards”
Conversely, the quality of life for nearly all humans will be considered miserable today compared to life in the 2030s-2040s, even compared to what was considered miserable 200 years ago, indicating fundamental material and existential improvements.
“the quality of life for almost all humans would be considered miserable today even 200 years ago”
Democracy has expanded globally: two centuries ago there was only one democracy; two and a half centuries ago there were no democracies; today about half the world lives in democratic countries, reflecting a broad trend toward democratic governance despite some countries remaining non-democratic.
“this has to do with democracy uh it's not perfect there places to where we don't have democracy about half the world lives in a democratic country but you know a little over two centuries ago there's only one democracy you go back two and a half centuries ago there was no democracies um there are players in the world that aren't Democratic but the number of democracies that's gone way up”
Ray Kurzweil is currently a principal researcher in AI and AI Visionary at Google.
“and of course now he's a principal researcher in AI Visionary at Google Ray”
Major technology companies including Google spend more time ensuring the truthfulness and safety of large language models than creating the technology itself, suggesting they prioritize accuracy and alignment over capability.
“looking at large companies not just Google but uh they actually spend more time uh trying to control it and maintain uh truth in what it says uh than in in creating the technology itself”
Kurzweil's new book contains 50 different graphs demonstrating progress in multiple dimensions of human wellbeing: declining poverty, longer life expectancy, increased income, improved literacy rates, and expanded education rates.
“in this new book which you'll be getting from us uh I've got 50 different graphs that show progress in every aspect of our well-being declining poverty longer life expect y increased income literacy education rates”