March 11, 2006
Aubrey de Grey Sees War On Aging Starting In 10 Years

Biogerontologist Aubrey de Grey says the highly goal directed engineering effort which he calls the coming "War On Aging" will begin in about 10 years.

Aubrey de Grey of Cambridge University, UK, has presented a cure for aging - Strategies for Engineered Negligible Senescence. The plan's focus is not to interfere with a person's metabolism, but to repair damage to the body over time, at the cellular level, rather than dealing with the aging process in its later stages.

"My point here is just that this is goal-directed rather than curiosity-driven," de Grey said. "I view medicine as a branch of engineering."

...

De Grey calls the time during which the technologies will experience the most development the War On Aging.

"I use the phrase to describe the period starting when we get results in the laboratory with mice that are impressive enough to make people realize that life extension is possible, and ending when the first effective therapies for humans are developed," de Grey said. "I estimate that the War On Aging will start 10 years from now, subject to funding of research, and will last for 15 years, but this latter estimate is extremely speculative."

When he refers to a point when the War On Aging ends my guess is that he's referring to the point where we have achieved the ability to extend life faster than the rate at which calendar clock time advances. From an article of his published in PLoS Biology Aubrey says if we can extend life expectancy in a year by more than a year's time then Aubrey calls that point "actuarial escape velocity" which is the point at which we can repair aging damage faster than it accumulates.

...that in which mortality rates fall so fast that people's remaining (not merely total) life expectancy increases with time. Is this unimaginably fast? Not at all: it is simply the ratio of the mortality rates at consecutive ages (in the same year) in the age range where most people die, which is only about 10% per year. I term this rate of reduction of age-specific mortality risk ‘actuarial escape velocity’ (AEV), because an individual's remaining life expectancy is affected by aging and by improvements in life-extending therapy in a way qualitatively very similar to how the remaining life expectancy of someone jumping off a cliff is affected by, respectively, gravity and upward jet propulsion (Figure 1).

The escape velocity cusp is closer than you might guess. Since we are already so long lived, even a 30% increase in healthy life span will give the first beneficiaries of rejuvenation therapies another 20 years—an eternity in science—to benefit from second-generation therapies that would give another 30%, and so on ad infinitum. Thus, if first-generation rejuvenation therapies were universally available and this progress in developing rejuvenation therapy could be indefinitely maintained, these advances would put us beyond AEV.

How can this be accomplished? Read about Aubrey's Strategies for Engineered Negligible Senescence (SENS) for achieving this goal. Also, Aubrey and Dave Gobel founded the Methuselah Mouse Prize to encourage scientists to extend the lives of laboratory mice.

The prize seeks to encourage development of technologies that will also extend human lives. But its most important effect will be in terms of how those advances come to be viewed by the general public. The sooner scientists extend the lives of lab animals the sooner the public will wake up to the feasibility of radically extending human lives. This realization on the part of the public will eventually lead to widespread public demand for the War On Aging. Anyone who donates to the Methuselah Mouse Prize is helping to make the War On Aging begin in earnest sooner rather than later. Anyone who promotes the message that ‘actuarial escape velocity’ (AEV) is achievable via SENS technologies within the lifetimes of most of the people alive today also is effectively arguing for the coming War On Aging.

Stop being a pacifist where death is concerned. Join the supporters of the War On Aging. Time to go into battle against the Grim Reaper.

Update: Jay Olshansky, Daniel Perry, Richard A. Miller, and Robert N. Butler, arguing for a more modest goal of decelerating the rate of aging say that the future costs of an aging population will increase so much that the costs of an accelerated pace of aging research are easy to justify in terms of potential future costs avoided.

Consider what is likely to happen if we don't. Take, for instance, the impact of just one age-related disorder, Alzheimer disease (AD). For no other reason than the inevitable shifting demographics, the number of Americans stricken with AD will rise from 4 million today to as many as 16 million by midcentury.4 This means that more people in the United States will have AD by 2050 than the entire current population of the Netherlands. Globally, AD prevalence is expected to rise to 45 million by 2050, with three of every four patients with AD living in a developing nation.5 The US economic toll is currently $80-$100 billion, but by 2050 more than $1 trillion will be spent annually on AD and related dementias. The impact of this single disease will be catastrophic, and this is just one example.

$1 trillion per year in future costs for Alzheimer's alone demonstrate the scale of the potential savings that could come from therapies to decelerate and even reverse aging. Already today's cost of diseases run into the trillions in health care costs plus additional even higher costs of lost productivity and strains on families and friends who help out the sick and invalid. Our spending on anti-aging research should be in the hundreds of billions of dollars per year.

Share |      Randall Parker, 2006 March 11 01:39 PM  Aging Debate


Comments
Nick said at March 11, 2006 5:36 PM:

It boggles my mind that we haven't caught on to this as a society. I think almost everyone still sees aging as a supernatural thing, that can't be understood or prevented/cured.

There was a meta-study just released, that projected that a 20% reduction in the disability rate for seniors, over the last 20 years, has already dramatically reduced projected costs of care.

100,000 deaths per day, due to aging! How can we let this go on?

I see Juvenon (see juvenon.com) and resveratrol as the first real aging reducers, and the first step in the series of aging reducers. Unfortunately, there doesn't seem to be a consensus on how to package resveratrol for effectiveness (except for wine). Anyone heard anything about that?

Curious George said at March 11, 2006 5:57 PM:

Excuse me Nick, but what do you mean "How can we let this go on?" It's aging for crying out loud...it's entropy-a law that all nature is following! We just get old and die-that's all. I feel like working to improve our lives while we are here is important, and yes, if we can lengthen our lives, that would be great. However, I feel like there would be better uses for the money.

Bob Badour said at March 11, 2006 6:15 PM:

Curious,

Polio is natural. Smallpox is natural. A whole host of sundry fevers and infections that led to perfectly natural mortality have been essentially eliminated. Infant mortality? Perfectly natural.

I am with Nick. We have a disease called senescence that kills millions causing incomprehensible suffering along the way. It has already killed three of my grandparents and perhaps some people you know and love. It robbed my maternal grandmother of most of the enjoyment of the final six years of her life while she cared for my grandfather.

How can we let this go on?

Randall Parker said at March 11, 2006 6:17 PM:

Resveratrol without wine: I was wondering the same question a few days ago while reading a description of what resveratrol does to metabolism. Anyway, yes, one commercial supplier looks like it might be packaging resveratrol in a way that retains some activity. See this posting by Reason on his Fight Aging blog for the details.

The company mentioned there is Longevinex. See this ImmInst discussion.

However, before getting too enthused about Longevinex as the only active resveratrol on the market see this later ImmInst discussion which includes a statement from the Life Extension Foundation about the biological activity of their resveratrol. Here's an LEF site search on resveratrol. I'd be inclined to go with the LEF offering.

Anyway, will a resveratrol supplement turn on Sir1? Will doing that allow you to live longer? One can only guess at this point.

Bob Badour said at March 11, 2006 6:17 PM:

P.S. Your talk of entropy sounds like you worry life extension will hasten the energy death of the universe. I suggest we should wait and cross that bridge when we get to it.

Kurt said at March 11, 2006 7:11 PM:

My daily regimen includes Resveratrol, CoQ-10, Carnosine (supposed to slow the formation of AGE crosslinks) and the Life-extension Mix. All of this comes from LEF. I highly recommend LEF supplements, even though they are somewhat more expensive than others. LEF is the best and they also finance some of their own anti-aging research as well. Saul Kent, the founder and president of LEF, is highly involved in cryonics as well.

More significantly, they are involved with something called the Timeship Project. This will be a new residential/industrial park development for biotech, nanotech, and other businesses related to life-extension and cryonics. It will also be a township for "life extension oriented" and other libertarian types analogous to the recently anounced town for Catholics that the founder of Domino's pizza is establishing.

Saul Kent has been active in life extension and cryonics for 35 years.

rsilvetz said at March 11, 2006 7:41 PM:

Normally I am the contrarian on here, however, this time I agree. Frankly Aubrey has done a good first step -- although he is Johnny-come-lately and riding a wave that has been building for 20 years. Lef.org has been around 20+years, neo-tech's Dr. Wallace identified biologic immortality as a moral imperative in 1986, hell, I was bitching about it in med school in 82-89 -- even had a hell of a fight with George Annas on the topic (and I hear he is no better on the stem cell issue -- geez how did he get tenure...). And heck, 1987 saw the publication of the great little tome If We Can Keep A Severed Head Alive... suggesting discorporation as a way of surmounting trauma and imminent death (lop head off and put on another organism -- and that has already been done several times).

This is not to sanction any kind of .gov project in the matter, which I would I believe at the present state of society would result in a decade of waste (ref - Genome Project -- took a privateer to make it happen and light a fire under .gov).

And of course, let's not forget that with the FDA/NIH axis of inertia, the sooner we blow away both organizations by total defunding of their operations, the better. We also need to get patents out of medicine as well... but these are topics for another time.

Randall Parker said at March 11, 2006 8:06 PM:

Robert Silvetz,

I disagree with your and others interpretation of the DNA sequencing being accelerated by commercial competition. The big advances were being made in instrumentation by private companies. The government, by shoving so much money at sequencing, gave those companies (ABI in particular) money to reinvest in successive generations of instruments. The acceleration of the rate of sequencing was going to happen anyway. Adding in Venter's technique sped things up some. But that was a one-off speed up that didn't even contribute to the sequencing of the hardest parts of the genome (big repetitive areas).

If we go back further into the 1980s then the NSF funding of CalTech guys to develop the earlier sequencing instruments was absolutely key and that work spun off into a private company. Also, next gen sequencers are going to feed off of discoveries being funded by government. I bet the same spin-off pattern is happening again. Make the basic discovery in academy. Then go visit VCs and get start-up money to commercialize it.

Yes, big money from government would speed up the development of SENS tech.

rsilvetz said at March 11, 2006 10:25 PM:

Hi Randall,

As we must often do -- we will simply agree to disagree. My final 2 cents and to you the last word:

Let me get this straight. .gov goes off and spends well over $3 billion (all costs, not just nominal budget fund), progresses at a snail pace and the OTA wants to kill it. Along comes Venter, private, at maybe $350 million total landed cost, to do the same, embarrasses them into action -- and you disagree with my interpretation? It might be time to think twice. Price and cost tell everything. Always.

Then there's Clinton's little stunt about trying to destroy Celera by declaring the genome unpatentable. (Nothing like changing the rules when .market has kicked .gov in the teeth.) So, if memory serves, the market only lost $50 billion that day... Do we really need a SENS effort from people like that? How many years did that market drop cost biotech? 2? 3? How many lives?

If that's not enough to think twice on, do a direct compare:

Computer market. Unregulated. Performance/Price always increasing. I have Moore's Law.

Healthcare. Regulated to death. Performance/Price sucks and is getting worse. I have the inverse of Moore's Law, ten years to get a drug that improves life expectancy of 10% of colon cancer patients at $cost of 17K per dose. Nothing really has changed in that field since I graduated in '89. With few exceptions everyone dies the same way by the same diseases in about the same time under mostly identical therapies. That progress for you! How many billions have .gov sank into NIH? CDC? FDA? HCFA? That's price/performance for you!

I am deploying/developing/clinically testing/marketing some nice paradigm-shifting tech in the next few years in the areas of oncology, nephrology and bioidentical hormone-replacement therapy. I'm down-in-the-trenches. This romanticized interpretation about how things happen as described above is pure poppycock. It doesn't happen that way. In fact, the academia to VC gridlock caused by the method patent and protein patent disasters is like LA highways on Friday afternoon -- nobody going anywhere at any speed.

But let me tell you -- the bulk of my costs are purely regulatory -- criminally high. The cost in dollars, time and lost patient lives makes me sick on a daily basis. Forgive me if I don't share the idea that the FDA/HCFA/CDC are of any NET-NET use -- and that a .gov SENS project would be helpful. I can imagine no greater tragedy than a .gov SENS project.

Thanks as always for the great resource of The Future Pundit.

James Bowery said at March 11, 2006 11:01 PM:

Let me get this straight. .gov goes off and spends well over $3 billion (all costs, not just nominal budget fund), progresses at a snail pace and the OTA wants to kill it. Along comes Venter, private, at maybe $350 million total landed cost, to do the same, embarrasses them into action -- and you disagree with my interpretation? It might be time to think twice. Price and cost tell everything. Always.

It was actually entirely proper to prevent the patenting of the human genome data. Technology is patentable but nature is not.

However, this isn't to say the market approach is inappropriate.

On the contrary, read this legislative proposal from 1989:

http://groups.google.com/group/sci.space/msg/ca694a60b0ed909b

For the enhancement of scientific knowlege and the required development of advanced technology, A National Science Trust shall be established, with funding authorized by Congress, for the purchase of information about the natural world from Eligible Parties (private entities owned and controlled by other such entities in the U.S. or its unified free- trade partners). No less than 2/3 of the components and services used by the Eligible Parties to acquire this information must be obtained from other Eligible Parties.

The National Academy of Sciences shall identify areas of scientific interest in which the quality of research results are quantifiable -- primarily in terms of information content. Examples of these kinds of research results are: DNA sequencing (human genome project), digital imaging of various phenomena (astronomical, planetary, terrestrial ozone-layer monitoring), quantitative behavior of systems in microgravity, quantitative mineral assay of various sites (terrestrial and nonterrestrial), etc.

A dollar amount, to be established in conjunction with Congress, shall be associated with each informative item and with varying degrees of accuracy of the information. That dollar amount will then be appropriated to The Trust to be paid out only in the event that an Eligible Party has delivered new information on the associated item of interest to a designated recipient. When a measurement has already been made, payout will be limited to information value corresponding to the increased confidence level of the measurement (e.g. additional significant bits or fractions thereof). In areas where an information flow is required (periodic sampling) the value of various sampling frequencies at the various degrees of accuracy (significant bits) will be included in the valuation of the measurement. Duplicate information flows will share the cash flow evenly. For superior information flows, the incremental increase in accuracy will enjoy less diluted access to funding flows allocated to those incremental increases in accuracy.

Income on The Trust will be used to adjust The Trust for inflation. Additional income from The Trust may be used to fund items within The Trust. In the event that an item is measured by a Party which is not an Eligible Party, and that information is available to the designated recipient -- the corresponding funding will be redistributed within The Trust. After-inflation losses will be redistributed within The Trust, deactivating items which are not currently being pursued by any Eligible Party.

Bristlecone said at March 12, 2006 4:55 AM:

I think this is a horrible idea.

You really want everyone to be immortal? Does "Overpopulation" mean anything to you? Do you really want to work forever?

Perhaps I don't understand the concept...exactly how long will people be living? And, do these people expect to retire at 65 and have subsequent generations support tham for hundreds of years?

If I don't understand, please educate me, but this sounds dreadful.

Bob Badour said at March 12, 2006 5:27 AM:
You really want everyone to be immortal?

Well, duh. Yes, of course, I do.


Does "Overpopulation" mean anything to you?

Intelligent people with relatively long lives do not now reproduce at replacement rates. I see no reason to expect an overpopulation problem.


Do you really want to work forever?

I expect to have some vacations. I might even retire for a few decades every century. But otherwise, yes, I want to challenge my mind and my body and feel like a productive, contributing member of society forever.

simon said at March 12, 2006 6:56 AM:

de Grey is interesting but it is hard for me to take him very seriously. While clearly a visionary, he represents the science fiction element of this important movement. Many of statements smack of whimsy rather than reason. His statement "I view medicine as a branch of engineering." is pure nonsense. Medicine is not anything like an engineering discipline. Medicine is not even a true science. We need serious scientists leading this charge if we are to really take on this challenge.

Randall Parker said at March 12, 2006 8:33 AM:

Robert Silvetz,

Let me get this straight. .gov goes off and spends well over $3 billion (all costs, not just nominal budget fund), progresses at a snail pace and the OTA wants to kill it. Along comes Venter, private, at maybe $350 million total landed cost, to do the same, embarrasses them into action -- and you disagree with my interpretation? It might be time to think twice. Price and cost tell everything. Always.

You misrepresent the facts.

1) Venter's costs were lower because he came later. The instruments from ABI went thru revs. Academic labs learned how to make them more efficient. Venter and the academic labs under contract to the government both became more productive as the project progressed primarily because the instruments advanced.

2) The government's money paid for more passes thru the genome than Celera did. Celera did like 2 maybe. The government funded several more. Those passes were needed to error correct.

3) The government - and not Celera - put a lot of effort into sequencing the parts of the genome that are hard to sequence.

4) The big advances in instrumentation first happened in academia, at CalTech in particular. Leroy Hood got funding from NSF which he used to take a mass spec developed for Mariner (again on the government's nickel) and improve it to make the first automated sequencer. I went to a talk he gave and heard this from him first hand back in 80 or 81.

4) Other CalTech guys in the 1980s - and still working on the government's nickel - solved important problems that led to their ability to go private and found a company to make a far better sequencer. They improved it over a succession of generations with most of their sales getting paid for by government money.

4) The company they founded - rather than Venter - made the bulk of the advances that drove down costs for both the government and Venter.

Venter's gotten himself great press. He made an important contribution. But the press has greatly exaggerated his contribution. Glyn Moody untangles some of the myths from the facts about the sequencing of the human genome in his book The Digital Code Of Life.

As for patents: This is another issue. But I do not think the sequence of natural genomes should be patentable.

Randall Parker said at March 12, 2006 8:36 AM:

Bristlecone,

If people do not age there will be no need for retirement and certainly taxpayers will not be expected to fund the retirement of people who have youthful minds and bodies.

Work forever: Well, I expect robots will do most of the work. But if you do not want to live you can always kill yourself.

Overpopulation: Stop people from having babies.

Nick said at March 12, 2006 9:42 AM:

"Overpopulation: Stop people from having babies."

Randall,

My understanding of the latest demographics research is that there is a very strong link between income/education, and reduced birth rates: almost all of the developed countries have lower than replacement birth rates, and in most of the developing countries birth rates are falling quickly (the few exceptions, like nigeria and saudi arabia, treat their women very badly, in part because oil income creates a very unequal income distribution, and supports authoritarian governments). My take on this is that no coercion is needed to prevent babies, just good government and economies, and that present trends indicate a peak in population in about 40 years, with declining population after that.

Does that make sense to you?

Randall Parker said at March 12, 2006 10:05 AM:

Nick,

Some countries have bottomed out on fertility and have started heading back up again. This is happening in a few African countries. Forget which ones.

My take is that Darwin wins out evenutally. People in Western countries who are having more kids have a stronger instinctual desire for kids. The next generation will have an even stronger desire, and so on. The fertility will drop only so far and then rebound.

Dr. X said at March 12, 2006 3:46 PM:

Overpop and immortallity: Birth rates tend to be tied in an inverse relationship to protien intake. But the people having big families these days tend to be of the 'fundamentalist' vien. Yes the meek shall inherit the earth.

On the other hand, we might be able to get off the planet for $10 a kilo in todays dollars, so there could be a mass exedus at some point.

Government funded research has its place: look at how much stuff came out of ww2 research.

aa2 said at March 12, 2006 4:09 PM:

Kurt - that timeship project sounds interesting. I share those ideas, and it would be nice to live in a community of like minded people.

aa2 said at March 12, 2006 4:19 PM:

Bristlecone - "You really want everyone to be immortal? Does "Overpopulation" mean anything to you? Do you really want to work forever?

Perhaps I don't understand the concept...exactly how long will people be living? And, do these people expect to retire at 65 and have subsequent generations support tham for hundreds of years?

If I don't understand, please educate me, but this sounds dreadful."


Thanks for taking the other side of the argument Bristlecone, so we can flesh out ideas. These are good questions that the public is going to ask. On overpopulation I personally don't believe we have that crisis in the developed world. If you look at a nation even like the netherlands they could easily double their population with today's technology. We can desalianate water, build tall buildings for people to live in, and we are increasing crop yields much faster then population is growing in the developed world.

With energy we could even make giant buildings to farm plants in, with many stories. On each story you have artificial lighting. Farming in a hydroponic style instead of the farming style of today. Of course worrying about that would only come to a country like the netherlands if there was a huge increase in population like 10 times. So at some point we might have to limit the amount of children people are having. My personal feeling is that as the life sciences advance to the point where we can stop aging, they will also be advancing in our ability to engineer children. So there won't be the same desire to procreate to sustain the race or pass on the good things in your person. Further pushing down the fertility in places like Europe.

aa2 said at March 12, 2006 4:28 PM:

"Do you really want to work forever? And, do these people expect to retire at 65 and have subsequent generations support tham for hundreds of years?"


For most of the citizenry their productive employment will be pointless by 2050 anyway. What is already happening to the automobile industry will happen across all industries as computers and robots rise in sophistication. What will be left is jobs in art/entertainment, engineering and science. And even those jobs will be phased out eventually.

So what will happen is the robots who replace the workers will support them. Something like a 30,000$ pension seems like a heavy burden to shoulder today for society and it is, but in the future it won't be. I suspect if people were returned to say the body and mind they had when they were 30, even if they were getting a pension they would work on top of that. In fact I already see that with people I know, the 55 year olds who have big pensions already, most still choose to work on top of that.

aa2 said at March 12, 2006 4:32 PM:

- "exactly how long will people be living?"

I think the goal is to have it so that people do not succumb to aging. Just as they by and large don't succumb to infections today. Of course I have no problem if some people don't want to live anymore, and take their own lives. And I'm not just saying that toungue and cheek, I believe in self-ownership, and part of that imo is the right to die.

Richard said at March 12, 2006 4:57 PM:

"giant buildings to farm plants in, with many stories. On each story you have artificial lighting. Farming in a hydroponic style instead of the farming style of today."

Kurzweil thinks we'll be synthesizing food via nanotech.

John Schloendorn said at March 12, 2006 7:25 PM:

Hmm, the story used to be that "a war on aging" would be inspired by robust mouse rejuvenation, and that it could happen in ten years, subject to massive funding for the mouse project. But massive funding for the mouse project wasn't there, which is why we've heard "in ten years" for the last five years. This time, there is no reference to the mouse project at all -- did anything change that I should know about?

Randall Parker said at March 12, 2006 7:30 PM:

John Schloendorn,

You should know that if you click thru on links you might find stuff beyond what I excerpted.

John Schloendorn said at March 12, 2006 7:55 PM:

Ahh, thanks ;-)

Joe Bain said at March 13, 2006 8:32 AM:

You can download or order a free documentary on current aging research at the Immortality Insitiute.

http://www.imminst.org/film.php

It is 1:45 long. If you can download the video via GoogleVideo or you can request a free one be sent to you.

Carl Shulman said at March 13, 2006 10:54 AM:

Randall,

What are your thoughts on artificial intelligence and technological singularities? You advocate spending tens of billions on energy, anti-aging research (which I agree are very valuable research topics), but don't seem to mention AI much, although strong AI could be used to quickly achieve the rest, and pure AI receives a fairly miniscule level of funding. Why not put a few billion into AI research?

Jim said at March 13, 2006 3:25 PM:

I find the statement "I view medicine as a branch of engineering." as a key concept that changes the game. VC's and entreprenuers have little business in science apart from commercializing key discoveries.

Engineering is different. it implies that you are taking advantage of the known science and designing a complete system rather than hoping to discover a statistically observable improvement due to an isolated compound from amongst a matrix of candidates. Engineering design becomes more technically powerful as the gov't funded research expands the scientific knowledge base of relevant biological mechanisms.

the body is ridiculously complex compared to any previous engineering undertaking, and entropy will prevent complete elimination of aging, just enable you to put the day off. something of a 'moore's law' for age expectancy.

Randall Parker said at March 13, 2006 4:56 PM:

Aside: I just upgraded my MovableType install to v3.2 and this has caused a few changes. Let me know if this causes any problems for any of you.

I'm going to be working on other changes and blogging will be lighter as a result.

Carl,

I'm afraid of AI. I do not want to be obsolesced by machines. I do not agree with those who think we'll easily avoid the bad outcomes.

I also think that biotech brings many serious threats as well. But advances in biotech to make us young again beat getting decrepit and dying from cancer.

Jim,

The curious thing about Moore's Law at this point is that it is getting harder to maintain it. Simple reason: rising nanojoules of energy needed per executed instruction. We've gotten to a scale where the electricity leaks because the distances between devices is so small.

BTW, this really will push AI research. Why? We need to do a lot more parallelism in computing or else performance increases will halt. What does great parallelism? Brains of biological organisms.

Brett Bellmore said at March 13, 2006 5:13 PM:

Jim, entropy needn't prevent a complete elimination of aging; Nothing in the laws of thermodynamics precludes a system staying organized indefinately, so long as it's got a source of energy and a heat sink.

This is not to say that we'll live forever, merely that people will be rather like radioactive isotopes, having a half-life, rather than a lifespan.

Carl Shulman said at March 13, 2006 5:22 PM:

I agree that the risks of bad outcomes are quite high, but I think that AI development is inevitable. AI-as-strategic weapon would make nuclear weapons like like flint arrowheads: would you prefer that the first humanlike, self-improving AI be developed by the Chinese government? Worse, by a project using evolutionary or unpredictable techniques that may result in an AI that eats the solar system for computing material? The longer we wait, the lower the barriers to entry, and the greater the likelihood that a project without the resources or desire to produce safe AI will be able to produce deadly dangerous AI.

Taking efforts like CYC and increasing their resources a hundredfold could allow them the time to take more precautions while moving faster than other projects, and increase the odds that AI would not be used to establish an eternal totalitarianism or exterminate most or all of humanity. If the options are to develop safe AI (with the cosmic benefits that entails) or to be killed by AI, I would suggest doing everything possible to shift the odds towards the former.

Randall Parker said at March 13, 2006 8:05 PM:

Carl,

You may be right. I do not know. Certainly the Chinese getting AI before us would be very bad news. But if that's the case we should stop letting Chinese come and work in our labs. Info about the advances is going to flow back and forth pretty quickly. How can we get significantly ahead of them and then stay ahead?

Carl Shulman said at March 14, 2006 4:02 AM:

The suggestion is for a Manhattan Project-type effort, both in terms of resources invested and in keeping interim results secret. As to getting "significantly ahead," from a strategic point of view self-improving AI is the endgame: even a lead of a few weeks or months could be utterly decisive. I'd be far more concerned about a screw-up in the AI's motivations (an existential risk), which is why I'd want to invest as much as possible. Increasing a lead in AI technology (in a classified project) would give breathing room to improve the odds of safe development.

Carl Shulman said at March 14, 2006 4:10 AM:

I would suggest a venture in the tradition of the Manhattan Project, both in terms of resources invested and in secrecy of interim results. DARPA already invests in AI research and has experience with keeping secrets: it just needs to scale up its efforts to reflect the fact that this is a technology that beats out nuclear or biological weapons as the ultimate WMD, and that development without sufficient precautions could cause the extinction of humanity.

Optum said at March 14, 2006 5:46 AM:

I suspect Western countries' fear of science (e.g. stem cell research, genetic engineering, anything having to do with the science of human nature) in comparison to more objective East Asian countries could make it inevitable that that region will eventually pull ahead of the West. Or maybe a national desire as that occurs to not fall behind would encourage our natural interest in science. Psychometrician Richard Lynn estimates embryo selection as a form of standard reproductive therapy would raise the average intelligence of the population by 15 IQ points in a single generation, and this gain could be repeated each generation until a theoretical maximum is reached after six or seven generations (this book received good reviews). If this were to occur, Western society can only fall so far behind before some of the sharper people start to emigrate.

remo williams said at March 15, 2006 6:50 PM:

Randall,
I'm not sure why you say Moore's Law is becoming harder to maintain. Is this measured in how many development hours are put into going the next step , or is it in terms of research spending? Obviously, it will be "hard" to reach the very limits of Moore's Law but by 2015 we should be well on our way to keep exponential power going in a new paradigm. I think it is possible we could say some lag period of 3 to 5 years where the trend stalls before taking off (maybe 2016 to 2020), but even that seems unlikely.

I've never understood why the Chinese getting AI "first" is a problem. That would transfer very , very quickly. If the US gets it "first" (quotes, because it may not be entirely obvious when it arrives), the Chinese -- and likely a more democratic China -- will have it soon after.

Randall Parker said at March 15, 2006 7:44 PM:

Remo,

The semiconductors are getting down to gate sizes and trace sizes that just do not leave many atoms between devices. They are hitting the wall. The problem is that too much electric leaks and the number of nanojoules per instruction has gotten out of hand.

The original Pentium in 1993 consumed about 13 nanojoules for instruction it executed. The Pentium Pro in 1995 nearly doubled performance, but consumed about 24 nanojoules per instruction.

The Pentium 4 in 2001 was about three times the performance of Pentium, but it consumed about 38 nanojoules per instruction, while the Pentium 4 had four times the performance but about 49 nanojoules per watt.

The Pentium M lineage of chips originally designed for mobile PCs changed the trend. The first model in 2003 had the same 12 nanojoules per instruction energy consumption as the original Pentium but about 2.7 times the performance. Core Duo, which came to market this year, has four times the performance and even lower per-instruction consumption of 10 nanojoules, Rattner said.

They are trying to increase performance by going for greater parallelism. They claim increases in performance that are probably exaggerations because so many apps not keep multiple threads busy.

The multi-core approach works better for servers that are handling many users. But it doesn't work as well for desktop apps. Also, some mathematical problems are hard to parallelize.

Their problem is that even if they could maintain 10 nanojoules per instruction that is not good enough. Every time the processor doubles in speed it still doubles in energy usage. They need to radically decrease energy usage per instruction executed. I do not see how they can do that.

What is needed are entirely different architectures. Most gates aren't doing anything during each instruction cycle aside from leaking electricity and therefore generating heat. Neural networks or other approaches are needed. But no one knows how to implement business logic or other problems in such architectures. How to program, debug, validate such designs?

Jim said at March 17, 2006 8:36 AM:

moore's law is largely based on reductions in distanced within the chip. obviously there is a hard physical limit below the atomic size, and certainly before that due to tunnelling effects.

there's an economic limit as well - moore's law has a corrolary for the exponential growth in the cost of a new chip fab. at some point the entire world market can support less than one fab.

moore's law is amazing how long it's held up, but some new approach to computing will put us on a new curve.

remo williams said at March 18, 2006 4:15 AM:

I agree that the end of Moore's Law which must occur in the next 10-12 years will put us on a new curve, but is it hard to imagine that new 3D techniques can't keep us on roughly the same 100 year exponential that a few including kurzweil have plotted? That is really what I meant about Moore's Law, as has been popularized, continueing on for decades.

In this sense , Kurzweil's Law o Accelerating Returns doesn't seem to be getting "harder and harder to maintain" if we talk about it in terms of a measurement like percentage of GDP.

Nick said at May 10, 2006 3:20 PM:

Here's a good article about population versus genetically determined fertility:

http://www.latimes.com/news/opinion/commentary/la-oe-barash10may10,0,7632432.story

"THE GERMAN PUBLIC was recently shocked to learn that 30% of "their" women are childless — the highest proportion of any country in the world. And this is not a result of infertility; it's intentional childlessness....People are inclined to eat when hungry, sleep when tired and have sex when aroused. But in most cases, we remain capable of declining, endowed as we are with that old bugaboo, free will. Moreover, when people indulge their biologically based inclinations, nearly always it is to satisfy an immediate itch, whose existence is itself an evolved strategy leading to some naturally selected payoff. A person doesn't typically eat, for example, with the goal of meeting her metabolic needs but to satisfy her hunger, which is a benevolent evolutionary trick that induces the food-deprived to help out their metabolism....

...For more than 99.99% of their evolutionary history, humans haven't had the luxury of deciding whether to reproduce: simply engaging in sex took care of that, just as eating solved the problem of nutrition. But then something quite wonderful arrived on the scene: birth control. Because of it, women (and men) can exercise choice and, if they wish, save themselves the pain, risk and inconvenience of childbearing and child-rearing, indulging themselves rather than their genetic posterity."

Putin today lamented the population decline in Russia.

I really don't think we have to worry about overpopulation in educated groups. The problem is the uneducated.

bi said at May 11, 2006 12:02 AM:

Jeez. Does anyone here understand that the idea of "replacement rate" is meaningless if death doesn't occur?

Nick said at May 11, 2006 8:43 AM:

bi,

Are you suggesting that population growth has to occur if death doesn't?

Let me clarify: if replacement rate is 50%, i.e. if every couple has 1 child, every generation is half the size of the preceding. If the population starts at 128, the next generation is 64, and 32, 16, and so on. The sume of that numerical series is 256 - that's an example of the general solution to that series, which is the total sum of the series is twice the starting value. So, the population doubles, and stops there. The same kind of thing applies if the rate any other number, as long as it's below 100%.

Now, realistically, there's still likely to be accidental death.

I would be delirious to have to deal with the problems caused by a reduced death rate. The worst that happens is that we all live in big condo buildings in dense cities. I'd rather live with that than die with the alternative.

christopher wojtasik said at July 23, 2009 1:32 PM:

Do you think that the latest discovery of the properties of rapamycyne in the life span of mice could be the big breaktrough that De Grey talks about? I mean, do you think, that he would consider the effects of rapamycyne on mice a "robust augmentation"? I don't remember the numbers right now, but from what i have in mind the achieved longer life expectancy was around 10 percent in middle-aged mice, so not very impressive. From the articles that i have read the big break was that the pill is the first drug ever found to elongate mammalian life span, somehow mirroring the known effects of calorie restriction. What do you make of all of that? Can the discovered effects of this substance be the beginning of serious research so that we will finally start gaining momentum on this and also, not mention, a geniune public interest in life exetension reaserch?

Randall Parker said at July 25, 2009 9:54 AM:

christopher wojtasik,

Aubrey de Grey is going after rejuvenation therapies that repair and replace aged tissue. At best rapamycin might slow down the rate of aging. Slowed aging means you still end up getting old.

Granted, slowing aging is valuable. Add another 10 years to your life and that increases the odds you'll still be around when the rejuvenation therapies that de Grey advocates will become available.

Post a comment
Comments:
Name (not anon or anonymous):
Email Address:
URL:
Remember info?

                       
Go Read More Posts On FuturePundit
Site Traffic Info
The contents of this site are copyright ©