Wednesday 30 November 2011

Pensions and Longevity

The current wave of strikes over pensions will come as no surprise to those in the corridors of power - and nor will the strikes achieve anything. The hard fact is, is that pensions as they now exist are unsustainable. The entire concept of a pension, is unsustainable.

The hard reality facing governments, is that life extension is just around the corner - and no government, on any side of the political spectrum, wants to get locked into pension payments that will essentially drain the public purse -  pensions are doing this already.
What we will see, incrementally, is the link between pension and salary will dissolve - and it will be replaced with a basic subsistence allowance. Nothing else will be affordable, if government is not to be hamstrung. Borrowing to pay pensions is not a sustainable policy. Private pensions, that rely on a lump sum to be invested in an annuity, will see their returns shrinking to such low levels , that they too become unviable. Most private pension funds are now insuring themselves against massive advances in longevity - which,when (it isn't if, it is when) this eventuates, will lead to the collapse of many private pension funds, which will not be in a position to keep paying out for decades longer than expected. Even a 5% increase in average longevity will cause a crisis. To judge by the percentages being achieved in lower animals, 5% is a very conservative figure - expecting a 20% increase, or more,  is probably not unreasonable.

Every week brings news of life extension achievements in various model organisms - only this week, an elegant genetic manipulation dramatically extended the lifespan of yeast, using a biochemical metabolic reaction chain known to exist in humans as well. Other studies indicate the drugs that promote mitochondrial fitness have a similar effect. And so the list goes on.

Policy makers are well aware that people currently receiving pensions could well survive longer in retirement as they did in employment. This is unsustainable. State pensions are paid out of tax receipts. The system was never designed for this. Ministers know it is unsustainable. Voters want to have their cake, and eat it. This cannot be, something has to give.  Most of the people striking today are unaware of the massive policy implications medical interventions are already having on life expectancy, and, as the saying goes, we ain't seen nothing yet.

The speed at which new compounds can be investigated, meaning the speed at which progress is being made, keeps increasing - more than exponentially. A major driver of change is the application of AI to chemistry. Novel compounds can then be investigated, also through automated, intelligent systems. Experiments that would have taken decades or months of painstaking labour, can now be streamlined and reduced to weeks, if not days, or, in some cases, seconds.

Society will need to adjust to these changes - those demonstrating today, in some ways, are not unlike the Luddites breaking the mechanical looms of industry - unable to come to terms with a paradigm shift, they rose up in futile rebellion.

Tuesday 15 November 2011

What kind of professions will most be at risk from software ?

Knowledge based and technocratic professions are most at risk from software. House painters, plumbers and such will eventually be in the firing line, but this will take longer, as hardware that can compete with the dexterity of a human body lags far behind software development, and AI development.


 As it is, a vast percentage of trading on the stock market floor is now carried out by computer agents. So are document searches in legal departments. Many of these applications, while removing people from employment, will simultaneously  will open up new opportunities, so that new types of job will spring up, as a result of the increased productivity.


Here is an example of the type of work that is currently being taken over by robots in labs the world over: Indeed, in many cases, reactions are carried out - modeled - inside the software itself - but for some very complex and poorly understood interactions, real life experimentation is still needed:




"one employee needs approximately two weeks to process 16 samples (of experimental glass recipes).
Researchers of the Fraunhofer Institute for Silicate Research ISC in Würzburg have developed a unit that carries out all these steps automatically. "It needs only 24 hours to process 16 samples," says Dr. Martin Kilo, manager of the expert group for glass and high-temperature materials at the ISC. "For this reason we are able to develop glass elements more cost-effectively than previously, by up to 50 percent." 

Another example is the massive surge in scholarship and productivity that is resulting from the application of AI to the fragmented documents of the Cairo Geniza, a huge repository of Jewish texts scattered across the globe, mostly in fragmentary form. On the one hand, researchers are being put out of a job, one that could have provided fodder for generations of academics. On the other hand,the raw material available to be analysed on a meta level has jumped exponentially - providing resources that will attract funding, and more employment of specialists. Similar progress has been made using software with the Herculaneum papyri, but here huge technical obstacles to reading the unrolled remaining scrolls have yet to be surmounted.

Monday 14 November 2011

Computers that can self-program and evolve

At present, computers can 'learn' to some degree, mainly by fine tuning algorithms, but they are not at present able to completely re-write their code, to the extent that they can alter their circuitry, and truly evolve into new architectures. To do that, a method would need to exist whereby software can recofigure the actual design of the chip itself as it learns.
As we head towards three dimensional architectures for chips, a new path has opened that appears to lead in this direction. The ramifications are massive. If a viable chip emerges that can be reconfigured autonomously by the software running it, then we can imagine the ability for computers to evolve and learn to be extended, by growing new architectures, and new software. 

When there are 100,000 channels

A friend of mine was joking about who would hold the remote, when there were 100,000 channels. I replied, no-one will be holding it, the remote will be interlinked to your brain.
A few posts back, I was talking about human-computer integration,and positing various possible scenarios. With alarming rapidity, one of these scenarios is now a reality - a viable interface that is not implanted, and has high resolution.
How long before we have equipment that allows us to access our smartphones,which are micro-computers, after all, not 'phones' at all....directly via our brains? How long before our brains can directly access wikipedia, for example? Or a dictionary? I think we are looking at quite a short time window.