Green, ethical, energy issues in the news

zeupater

Martyn1981 said:

Not good news, but also no surprise, as we slow walk the shift away from FF's.

‘World heading for 2.5°C global warming’

If transformative action is not taken now, the Paris Agreement goal target will be missed, according to Wood Mackenzie.

The world is on a 2.5°C warming trajectory according to Wood Mackenzie’s Energy Transition Outlook report.

If transformative action is not taken now, the Paris Agreement goal to limit the average temperature increase to below 1.5°C will likely be missed, concluded the report.

Low carbon power supply and infrastructure needs to scale up at twice the pace built in the last decade – made more difficult by the current delays faced by renewables assets due to limited grid interconnections.

Hi

The caveat also being 'no surprise' because the report authors are contracted almost exclusively by corporations with various vested interests .... tell them the narrative and they'll likely write a report to support it ...

Just a couple of hours ago I mentioned that the COP27 conference effectively reported the same conclusion in November last year, so this report isn't really anything newsworthy, likely just being a nudge in the direction of those holding the purse strings, of course, that being on behalf of their sponsoring customers.

Anyway, what does anyone really expect? .... has everyone missed the fact that almost the entire globe was locked down for around 30% of the intervening time between the Paris agreement and the COP27 finding that almost everyone involved was behind target? ... as such it's almost incredulous that the very same governments & institutions that agreed to meet Paris accord targets were the ones vigorously pushing for extreme lockdown, and are now pushing the narrative that more needs to be done and that needs to be done faster to meet what are effectively arbitrary targets against an arbitrary calendar ... utter hypocrisy, especially considering the emissions reductions that should have accrued over the various lockdowns and the related reduction in economic activity which even continues to date ...

Targets are set and agreed according to what is considered possible, that's in physical scale using an allocated resource within an available time ... as anyone who has the most fundamental understanding of project planning knows, you cannot simply change any one aspect without impacting elsewhere ... if the available time is reduced because of a couple of years of lockdown, the target cannot be met ... this resulting in acceptance or some form of recovery action likely more money to support more resource to deliver the equivalent outcome within a shorter timescale, the caveat, of course, being that throwing resource at the problem is in no way limited by other factors, such as existing manufacturing capacity and the timescales involved in expanding any bottleneck capacity .... 

The logical approach would be to accept that any 2030 targets be indexed by the length of the delay's root cause (pandemic related lockdowns), so we're probably looking at 2032, the overall excess net emissions resulting from the delay should be quantified, with the ensuing emission mitigation being added to the target reduction for the 5 year period starting in 2030 as described and allowed for in the original 2015 Paris Accord .... 

Logic?, who uses that when it's possible to leverage a situation for financial gain! ... don't worry, just make the consumer & taxpayer pay more !!!

HTH ... Z (Potential dictator in waiting ... if only!!!)

zeupater

Grumpy_chap said:

zeupater said:

the basis of income tax (based on the fair exchange of labour for monetary gain) is rapidly replaced by robots that build robots to replace almost all human manual & administrational economic activity, then how does the economy work? .... as value is effectively based on the accumulated quantity of labour to produce a product, what happens to costs when there's little or no labour involved in the creation of that product? .... 

I would not be worried about that just now, and I doubt that investors are considering that as a serious impact either.

We have been promised or threatened, depending upon your point of view, that everyone will be made redundant by machines "within the next decade" and all we will have to do is enjoy endless leisure time since I was at primary school.  I am not convinced that this promise or threat is anywhere nearer now than it was then.  If it's ever going to happen, it had jolly well be along fairly pronto as I will feel pretty hard done by if everyone gets to stop work and enjoy endless leisure time just about the same time as I reach retirement, having worked through until that point 

Hi

The issue is that what would currently be considered as industrial robots which are built to specification/order and programmed to perform specific tasks at considerable cost aren't the same as the AI equipped general use robots which are programmed to learn how to perform simply by observing repetitive example, which is effectively the same way the animal kingdom does it ... a big cat learns how to hunt by simply observing other big cats hunting!!

The scale of change in the pipeline is nothing like the automation of automotive plant in the 1980s, that's relatively child's play ... what's coming will come quickly and will act as a catalyst to change almost every industry, especially those that have historically been practically impossible to employ automation at an economic level .... think about a scale ranging from unskilled labour harvesting leaks & onions at one end to something like hospital surgery at the other, all being performed by the same construct of robot, maybe even the same robot at different times in it's useful time-span ... 

It's no longer sci-fi, it's effectively here ... have you seen the latest iteration of Tesla FSD (v12)? - that's a completely new software stack using 'learn by example' AI .... teach it how the best drivers drive and it not only has the ability to outperform the average driver on any metric, it has the learned skillset to not only match the best, but due to the faster processing capacity and ability to measure far more accurately can logically better it's own example based teachers .... once the final release candidate is ready for mass use, that's effectively it and the future is here because the software stack running the AI robot on wheels is effectively the one which can run any other form of robotic AI device which benefits from a learn, process and perform approach, including human form variants if necessary ... and why not as almost the entire planet's infrastructure is built around and to accommodate the human form, so why add unnecessary cost & complication where it's not needed!! ... 

As originally mentioned, the timescales are much shorter than many currently envisage and governments & corporates all over the world are finally waking up to what the potential impacts of employing this level of AI based automation can be ... there's even been AI briefing to US government committees over the past few days, okay they're more concerned with security & control at the moment, but it shows that it's high on the agenda ...

HTH - Z (I fancy a butler to bring my G&T ... >"Jeeves, get your skinny actuators in here PDQ"! ... <"Brain the size of a planet and all I do is pour drinks!")     

michaels

zeupater said:

Grumpy_chap said:

zeupater said:

the basis of income tax (based on the fair exchange of labour for monetary gain) is rapidly replaced by robots that build robots to replace almost all human manual & administrational economic activity, then how does the economy work? .... as value is effectively based on the accumulated quantity of labour to produce a product, what happens to costs when there's little or no labour involved in the creation of that product? .... 

I would not be worried about that just now, and I doubt that investors are considering that as a serious impact either.

We have been promised or threatened, depending upon your point of view, that everyone will be made redundant by machines "within the next decade" and all we will have to do is enjoy endless leisure time since I was at primary school.  I am not convinced that this promise or threat is anywhere nearer now than it was then.  If it's ever going to happen, it had jolly well be along fairly pronto as I will feel pretty hard done by if everyone gets to stop work and enjoy endless leisure time just about the same time as I reach retirement, having worked through until that point 

Hi

The issue is that what would currently be considered as industrial robots which are built to specification/order and programmed to perform specific tasks at considerable cost aren't the same as the AI equipped general use robots which are programmed to learn how to perform simply by observing repetitive example, which is effectively the same way the animal kingdom does it ... a big cat learns how to hunt by simply observing other big cats hunting!!

The scale of change in the pipeline is nothing like the automation of automotive plant in the 1980s, that's relatively child's play ... what's coming will come quickly and will act as a catalyst to change almost every industry, especially those that have historically been practically impossible to employ automation at an economic level .... think about a scale ranging from unskilled labour harvesting leaks & onions at one end to something like hospital surgery at the other, all being performed by the same construct of robot, maybe even the same robot at different times in it's useful time-span ... 

It's no longer sci-fi, it's effectively here ... have you seen the latest iteration of Tesla FSD (v12)? - that's a completely new software stack using 'learn by example' AI .... teach it how the best drivers drive and it not only has the ability to outperform the average driver on any metric, it has the learned skillset to not only match the best, but due to the faster processing capacity and ability to measure far more accurately can logically better it's own example based teachers .... once the final release candidate is ready for mass use, that's effectively it and the future is here because the software stack running the AI robot on wheels is effectively the one which can run any other form of robotic AI device which benefits from a learn, process and perform approach, including human form variants if necessary ... and why not as almost the entire planet's infrastructure is built around and to accommodate the human form, so why add unnecessary cost & complication where it's not needed!! ... 

As originally mentioned, the timescales are much shorter than many currently envisage and governments & corporates all over the world are finally waking up to what the potential impacts of employing this level of AI based automation can be ... there's even been AI briefing to US government committees over the past few days, okay they're more concerned with security & control at the moment, but it shows that it's high on the agenda ...

HTH - Z (I fancy a butler to bring my G&T ... >"Jeeves, get your skinny actuators in here PDQ"! ... <"Brain the size of a planet and all I do is pour drinks!")     

Sadly the problem is that such systems are black box, we don't know how they are making their decisions or what the consequences might be.

Start with chat gpt, sometimes it introduces random made up 'facts' into its answers.  Supposedly we don't know why.  MY guess is because it uses 'training data' from stuff humans have written - humans who sometimes introduce made up facts (or misremembered facts or facts that whilst based n reality have been altered enough in the retelling to not immediately match with the source).  Thing is it can be very useful but not necessary reliable in critical situations.

Now look at financial trading, AI must be better at maximising profits than humans who can't process nearly as much data, nearly as quickly.  But perhaps AI will learn that it can manipulate the markets using particular trading strategies that result in suboptimal outcomes over all.

And how about energy trading, AI can be used so that distributed assets such as batteries can generate income by providing 'grid services' when supply and demand go out of balance - but a black box profit maximising algorithm might start to deliberately provoke imbalance in order to profit from it - not because this is what its human creators had in mind but because its algorithm merely tries to follow its 'maximise profit' goal.

And now how about battlefields, loads of data requiring real time reaction, AI must beat human decision making but again the method it might use to reach its 'goal' is not known and may be unacceptable.

People think AI is just fluffy nice stuff like chat gpt which 'can't really hurt anyone'.  We are in for a rude shock.

thevilla

Who needs AI to cook the books? Decades ago Enron manipulated power outages in the US for profit. There's a strong suggestion the UK energy market was manipulated over the winter to increase prices by reducing bids for generation.  The UK one isn't proven but who would be surprised?

zeupater

michaels said:

zeupater said:

Grumpy_chap said:

zeupater said:

the basis of income tax (based on the fair exchange of labour for monetary gain) is rapidly replaced by robots that build robots to replace almost all human manual & administrational economic activity, then how does the economy work? .... as value is effectively based on the accumulated quantity of labour to produce a product, what happens to costs when there's little or no labour involved in the creation of that product? .... 

I would not be worried about that just now, and I doubt that investors are considering that as a serious impact either.

We have been promised or threatened, depending upon your point of view, that everyone will be made redundant by machines "within the next decade" and all we will have to do is enjoy endless leisure time since I was at primary school.  I am not convinced that this promise or threat is anywhere nearer now than it was then.  If it's ever going to happen, it had jolly well be along fairly pronto as I will feel pretty hard done by if everyone gets to stop work and enjoy endless leisure time just about the same time as I reach retirement, having worked through until that point 

Hi

The issue is that what would currently be considered as industrial robots which are built to specification/order and programmed to perform specific tasks at considerable cost aren't the same as the AI equipped general use robots which are programmed to learn how to perform simply by observing repetitive example, which is effectively the same way the animal kingdom does it ... a big cat learns how to hunt by simply observing other big cats hunting!!

The scale of change in the pipeline is nothing like the automation of automotive plant in the 1980s, that's relatively child's play ... what's coming will come quickly and will act as a catalyst to change almost every industry, especially those that have historically been practically impossible to employ automation at an economic level .... think about a scale ranging from unskilled labour harvesting leaks & onions at one end to something like hospital surgery at the other, all being performed by the same construct of robot, maybe even the same robot at different times in it's useful time-span ... 

It's no longer sci-fi, it's effectively here ... have you seen the latest iteration of Tesla FSD (v12)? - that's a completely new software stack using 'learn by example' AI .... teach it how the best drivers drive and it not only has the ability to outperform the average driver on any metric, it has the learned skillset to not only match the best, but due to the faster processing capacity and ability to measure far more accurately can logically better it's own example based teachers .... once the final release candidate is ready for mass use, that's effectively it and the future is here because the software stack running the AI robot on wheels is effectively the one which can run any other form of robotic AI device which benefits from a learn, process and perform approach, including human form variants if necessary ... and why not as almost the entire planet's infrastructure is built around and to accommodate the human form, so why add unnecessary cost & complication where it's not needed!! ... 

As originally mentioned, the timescales are much shorter than many currently envisage and governments & corporates all over the world are finally waking up to what the potential impacts of employing this level of AI based automation can be ... there's even been AI briefing to US government committees over the past few days, okay they're more concerned with security & control at the moment, but it shows that it's high on the agenda ...

HTH - Z (I fancy a butler to bring my G&T ... >"Jeeves, get your skinny actuators in here PDQ"! ... <"Brain the size of a planet and all I do is pour drinks!")     

Sadly the problem is that such systems are black box, we don't know how they are making their decisions or what the consequences might be.

Start with chat gpt, sometimes it introduces random made up 'facts' into its answers.  Supposedly we don't know why.  MY guess is because it uses 'training data' from stuff humans have written - humans who sometimes introduce made up facts (or misremembered facts or facts that whilst based n reality have been altered enough in the retelling to not immediately match with the source).  Thing is it can be very useful but not necessary reliable in critical situations.

Now look at financial trading, AI must be better at maximising profits than humans who can't process nearly as much data, nearly as quickly.  But perhaps AI will learn that it can manipulate the markets using particular trading strategies that result in suboptimal outcomes over all.

And how about energy trading, AI can be used so that distributed assets such as batteries can generate income by providing 'grid services' when supply and demand go out of balance - but a black box profit maximising algorithm might start to deliberately provoke imbalance in order to profit from it - not because this is what its human creators had in mind but because its algorithm merely tries to follow its 'maximise profit' goal.

And now how about battlefields, loads of data requiring real time reaction, AI must beat human decision making but again the method it might use to reach its 'goal' is not known and may be unacceptable.

People think AI is just fluffy nice stuff like chat gpt which 'can't really hurt anyone'.  We are in for a rude shock.

Hi

Totally agree, and that's why the cautionary position is being taken by the more adult leaders in the industry and why they're currently attempting to have some form of regulatory body to be formed in order to mitigate potential risk arising from corner cutting and other irresponsible actions as you rightly illustrate above...

The issue with the term AI is that although it exists in multiple forms, they're all grouped together in one bucket. Many forms of what is considered AI are simply decision branching algorithms which are hard coded ... they simply look for patterns as per code written by individuals and when the software makes a mistake the solution is to add additional subroutines consisting of new lines of code - as such this isn't AI in it's true form, it's simply a computer program specifically designed to perform in the way it's been specifically written to perform. Decades ago I had experience in developing & implementing AI type solutions utilised in industrial processes which did exactly this ... they were (and still are!) classified as AI, but they weren't, they were only as intelligent as they had been coded to be and couldn't 'learn' anything outside the confines of the limited knowledge of those that designed the workflow and wrote the code. I'm pretty sure that many of the issues that many raise are based on this level of understanding of AI, but we're now on the brink of a step-change in capabilities when you can visually teach AI what to do as well as it can be done by the best humans in that field (eg F1 racing drivers or best operators of an industrial machine) and have it repeat what it's learned perfectly, not only to the level of the best, but taking on board the small efficiency improvements learned from each observed example. Just consider that some F1 drivers perform better than 'the best' on different parts of the track or in different conditions ... that's where true AI comes into it's own, it self learns what examples to select and mimic to meet it's required goal ....

The problem then is the parameters for setting of the goal and the quality of the initial learning datasets ... allow poor data to be mistakenly injected into the learning 'curriculum' by a mistaken teacher and you'll get a mis-educated class of kids ... worse still ... intentionally inject mistaken data into the learning 'curriculum' and you'll be on the way to controlling the current (& future!) thought process of the very same class of kids - it's been done many times throughout history in order to gain levels of power & control, so that's why the cautionary position of having an AI regulatory body and underlying standards is hugely important, especially when the outcome resulting from potential failure is considered unacceptable ... 

So what is unacceptable?, well that's a fluid concept on a wide ranging scale ... think at one end of the scale you have financial collapse or nuclear war and at the other end there's a single unharvested spring onion left in a 5 acre field. Somewhere along the scale there's a car accident involving a pedestrian fatality .... is that acceptable?, should we ban the use of AI because of the fatality? ... even if the accident wasn't the pedestrians fault? ... Okay, then what happens to the proportion of the current 1.3 million annual road fatalities (~50% of which are pedestrians) which were likely to have been saved if the AI was being used ... weighing that one fatality against literally somewhere between hundreds of thousands and >1.3 million still means that there's +1 on the one side of the equation, but lots more on the other .... acceptable? ... I think so - and so should everyone else, probably even the unfortunate individual(s) who would suffer the fate of the pedestrian used in the example .... risk is always a factor, but not taking steps to mitigate the overall level of risk because the solution still contains some level risk is effectively insane! ...          

HTH - Z (Can the process of risk assessment & mitigation be inherently risky ?) ...

Martyn1981

thevilla said:

Who needs AI to cook the books? Decades ago Enron manipulated power outages in the US for profit. There's a strong suggestion the UK energy market was manipulated over the winter to increase prices by reducing bids for generation.  The UK one isn't proven but who would be surprised?

Similar thing in Australia about 4yrs ago, with the gas generators not maxing their production when demand peaked, which created a shortfall, and spike prices/profits, that earned a lot more, than had they generated at max.

Good news though, is that that was the spark that lit the race to battery deployments, large and small.

Coastalwatch

Something we on here are already well aware of but thought it worth posting in case any doubters are still not convinced.

Research shows heat pumps perform better than fossil fuel heating in mild cold climates

A British research group has aggregated information from seven field studies on heat pumps from around the world and has found air-source devices have an average coefficient of performance (COP) of 2.74 when temperatures are above −10 C. Below that, COP is between 1.5 and 2.

https://www.pv-magazine.com/2023/09/15/research-shows-heat-pumps-perform-better-than-fossil-fuel-heating-in-mild-cold-climates/

michaels

Coastalwatch said:

Something we on here are already well aware of but thought it worth posting in case any doubters are still not convinced.

Research shows heat pumps perform better than fossil fuel heating in mild cold climates

A British research group has aggregated information from seven field studies on heat pumps from around the world and has found air-source devices have an average coefficient of performance (COP) of 2.74 when temperatures are above −10 C. Below that, COP is between 1.5 and 2.

https://www.pv-magazine.com/2023/09/15/research-shows-heat-pumps-perform-better-than-fossil-fuel-heating-in-mild-cold-climates/

To think there was probably a research grant to pay for that research....

Martyn1981

This one was new news to me, a new form of generation for slow water generation. A bit like a waterwheel.

Rivtide seals turbine prototype grant

Rivtide Power has been awarded a £100,000 SMART: SCOTLAND feasibility study grant by Scottish Enterprise to develop a Mark 2 prototype of its Mass of Water Turbine (MOWT).

The system is a hydrokinetic turbine that effectively generates power from slow-flowing water such as rivers, tidal estuaries and ocean currents.

The patent0protected technology is entirely scalable from low kilowatts to multi-megawatts said the company, a subsidiary of MWNW Consulting.

On Googling to get more info, I found almost nothing, but this paper explains the principle. So best of luck to them.

MOWT – Mass of Water Turbine

Heedtheadvice

I am a great believer in trying to maximise things Iike tidal flow energy production as long as the environmental impact is not too detrimental and the cost gives value.

Predictability is one of the big pluses.

Like everything not as simple and straightforward as it might be hoped with several schemes not yet progressing far - except of course traditional hydro and pumped storage.

Ideas and research such as linked to above should be welcomed if only for potential diversification or addition to current 'green' energy sources.

As an experienced engineer working on many a project I was dissapointed by Zeupeters comments on time slippage being the only result of the delay on the climate zero aim.

Adverse things such as a delay often trigger rethinks in methods to achieve the aim and still keep to time, quality and cost. Mind you some of those do seem to have disappeared in the management of some projects! 

We need to rise to the challengeof net zero ( based as they are on our best system modeling rather than absolute certainties) . We cannot afford to fail to stop global warming timely and that needs to be the number one priority. Other things may be difficult to accept, costly or politically destructive but pale into insignificance by comparison.

We are all being slow to learn and this year's events to date should underline the need.