I wondered what it would look like with if the prompt didn't include an assumption there will be nuclear power and that it wouldn't be subsidised either.
Your results are interesting. It minimizes batteries and while it says add 400MW of pumped hydro, it looks like that backup is additional transmission lines.
I think the big thing you missed was more criteria about performing deep research, asking for citations, etc. I've found the boilerplate I add to my prompts for parts like that make a big difference.
This is a poor solution. It has gas at 50% for low winds/solar. It talks about GET which will improve the existing transmission lines, but no batteries, no additional pumped hydro, and no mention of nukes even though it was ruled out.
I think you (and I) found a prompt that has it struggling to fit a solution. And it's solution assumes that wind/solar won't go to zero for any significant time. So yes, there's limits to the utility of this. You don't want to overfit but you do need to give appropriate guidance.
Of course what constitutes "appropriate guidance" is going to become the new argument.
I then threw it at Gemini - https://g.co/gemini/share/395ebc8e7aa3 and that is a good solution. So you definitely want to use several A.I.s and compare. And where they're different, dive in.
Thank you, I was unsurprised by the analysis presented, but completely taken aback by the 'competency' and succinctness of the AI output. It really is a game changer.
Can you ask about what it would take to avoid the delays and cost overruns of the Vogtle reactors in Georgia? Or what the costs would be if Colorado's experience mirrors Georgia?
If I was doing this I first would talk to the Korean contractor who has a great track record of delivering their plants on time and at budget. Second I would talk to the contractors on the Vogtle plant because the last plant was 30% less than the previous. Find out if they learned even more that could be applied to the next job.
I wondered what it would look like with if the prompt didn't include an assumption there will be nuclear power and that it wouldn't be subsidised either.
https://www.perplexity.ai/search/you-are-an-expert-on-the-power-PaJjX_R1TBC9szayKeJlcQ
Your results are interesting. It minimizes batteries and while it says add 400MW of pumped hydro, it looks like that backup is additional transmission lines.
I think the big thing you missed was more criteria about performing deep research, asking for citations, etc. I've found the boilerplate I add to my prompts for parts like that make a big difference.
I did my prompt removing the nuclear parts - https://www.perplexity.ai/search/you-are-an-expert-on-the-power-rGb1mIPDT_2mFzjx3LE4fA and got a bit different mix where 50% of our generation is renewables.
The prompt is everything. On looking at the one you got and my recent one, I would rework that to say assume no new transmission lines to other ISO/BA regions. Because Germany has shown that to be a disaster. That gives: https://www.perplexity.ai/search/you-are-an-expert-on-the-power-rGb1mIPDT_2mFzjx3LE4fA
This is a poor solution. It has gas at 50% for low winds/solar. It talks about GET which will improve the existing transmission lines, but no batteries, no additional pumped hydro, and no mention of nukes even though it was ruled out.
I think you (and I) found a prompt that has it struggling to fit a solution. And it's solution assumes that wind/solar won't go to zero for any significant time. So yes, there's limits to the utility of this. You don't want to overfit but you do need to give appropriate guidance.
Of course what constitutes "appropriate guidance" is going to become the new argument.
I then threw it at Gemini - https://g.co/gemini/share/395ebc8e7aa3 and that is a good solution. So you definitely want to use several A.I.s and compare. And where they're different, dive in.
Anyways, good job on what you tried.
Thank you, I was unsurprised by the analysis presented, but completely taken aback by the 'competency' and succinctness of the AI output. It really is a game changer.
Up your alley ... a NON-AI generated energy plan for a different situation:
Japan targets renewable energy as main power source by FY2040
https://www3.nhk.or.jp/nhkworld/en/news/20250218_B03/
I enjoyed your post.
I haven’t had the time to compare these reports with the liftoff report by the DOE but my initial impression is that it very similar.
I believe this is the “amended”
https://liftoff.energy.gov/wp-content/uploads/2024/10/LIFTOFF_DOE_AdvNuclear-vX7.pdf
Keep up the good work!
The link is bad - can you fix it?
thanks - dave
Sure!
From this website you can get the full report, fact sheet, and summary.
https://liftoff.energy.gov/advanced-nuclear-2/
Interesting --
Can you ask about what it would take to avoid the delays and cost overruns of the Vogtle reactors in Georgia? Or what the costs would be if Colorado's experience mirrors Georgia?
If I was doing this I first would talk to the Korean contractor who has a great track record of delivering their plants on time and at budget. Second I would talk to the contractors on the Vogtle plant because the last plant was 30% less than the previous. Find out if they learned even more that could be applied to the next job.