1minusp

Does this allow for a predefined budget?

Even if zero net new insight, he has access to information on most things, not just AI, that the layperson might never have, even in the future. Not intended as praise at all (i'd say this is true for a lot of C-level in a lot of big companies).

Bene-gesserit have entered the chat!

I generally agree with this comment, but what option does a decision maker have here? (apart from similar products that probably will end up doing the same things anyway). Are there equivalent scale/functionality products that can truly serve as an option?

I'd like to be able to do something similar, but the old batteries in these things seem like a point of catastrophic failure. How is that dealt with?

Feels to me like the toolchain for using LLMs in various tasks is still in flux (i interpret all of this as "stuff in different places like .md or skills or elsewhere that is appended to the context window" (i hope that is correct)). Shouldnt this overall process be standardized/automated? That is, use some self-reflection to figure out patterns that are then dumped into the optimal place, like a .md file or a skill?

I'd love to believe this, but very recent history has shown (in the US at least) that we are moving backwards and trying to resist renewable energy.

Can be argued that there is intuitive satisfaction/pleasure/utility that spectators gain from watching sports competitions. The payoff is a lot more obvious/instant. Whereas with a lot of tech these days, what needle are we really moving? Are people truly happier scrolling for two hours, compared with watching an edge-of-seat soccer game?

I recall way back in the day when Stephen Fry used to host QI, they did a bit about a lasagne battery. Sean Lock was on the panel if i remember and spun it out into a ipod-style "lasagne-pod"

i think the applications to spy-craft could be quite interesting here. Something for the next mission impossible movie maybe?

I thought this was about scots/scottish dialects.

I find this interesting but the nomenclature escapes me: How is T(z) = z + zT + zT^2 + ... ? I find the jump from the functional programming description of the tree to this recurrence relation not intuitive

This. There seems to be one of these announcements every so often, and i havent seen any of them used at scale, or making any kind of dent in the status quo.

high fidelity holograms of people for remote presence

So GFS->graphcast = more accurate coarse-grained prediction (0,25 degree iirc), -> WRF for fine grained interpolation of forecast?

Do you use WRF as an input? I think graphcast uses some kind of NWP input (qhich in my mind i've equated to WRF). Thank you for offering to answer questions!

Any comments from experts on how much better these models are compared to running base WRF over GFS forecasts?

What are some good metrics to evaluate LLM output performance in general? Or is it too hard to quantify at this stage (or not understood well enough). Perhaps the latter, or else those could be in the loss function itself..

There an interesting line of thinking i've heard (apologies, i forgot which podcast i heard it on), that states that forecasts of electric load growth due to AI are overblown wildly, since they are based on very recent data that only reflects the early-stage explosive growth, and neglects to consider eventual efficiencies that tend to get built into such activities. Electrification of other more 'boring' areas like transport, heating/cooling etc are order of magnitude higher at least.

Hardly worth ones salt

Cost of undergrounding power lines is large. Especially over the distribution network (in that there is a lot more of it over space)

Large drilling costs from where I am in the south due to ground being mostly (lime?)stone: makes it cost-infeasbile. Other areas might have it easier though.

this is neat! I'll have to try this out.

Will be nice to be able to upload a picture/sketch instead of having to sketch something out in the scratchpad. I tend to sketch block diagrams from time to time and hate having to use some tool to draw/drag/align things. Very cool idea though!

Personal data point: I've had to move out of the bay area over a year ago after I failed to purchase a place due to the cost and competition for housing (tried hard, made multiple competitive offers, narrowly missed etc). Still hurts, its such a gorgeous part of the world to be. I doubt i'll ever be able to move back, most of the places we bid on are up several 100Ks over just a year.

Kind of Okey: Moved away from SV because we got sick of being beaten out on offers for homes (over a dozen attempts). Slightly better financially and the family is still together so perhaps i should be more thankful. Continued struggle with existing job and not being able to break into a role/company that will be more fulfilling or financially rewarding. I get stuck in this loop where i see people on linkedin or acwuaintances get jobs that i think i like, and icarry too much tension and fear of failure into interviews i guess.

any experts who can verify the validity of the claim of being low cost?

I work in a closely related space: parent's comments are accurate. Some of these items can be solved with better forecasting models, but the issue is that almost no-one actually cares about 'better' forecasting at inidividual meter levels. Sysytem-wide (or at least substation level) forecasts are well studied for supply/demand considerations (at a minimum). Also, the variability in consumption patterns at an individual level are large. IOW, one might be able to generate forecasts accurately for 'typical' residential consumption (at a per-household level), but commercials can be very different. Ideally, IF (and this is a BIG if), forecasts for "most" premises was accurate enough, then one could make the claim that the missing data shouldnt really cause a recalculation of bill once the actual consumption data comes thru, except in cases where the error is large. Guess that means that the error check/correct process needs to continue to exist.

The meter being replaced shouldnt be a major issue, this is relational (at least we treat this all as relational) data that should be captured (by the customer information 'CIS' system), and should be available with a 1-2 day delay. Similar argument applies to other relational aspects of the premise under consideration. Not saying that those are easy (more ways for this process to have gaps).

Know that parents comment is /s but in truth, it really feels like interpretations of body language and text to indicate hawkishness/dovishness actually seems to matter.

Share, yes absolutely. But also, keep in perspective, that these are ~$500K+ jobs and the (certainly smart) folks in these jobs can also move to other parts of google that arent as hard to deal with. So, it is understandable that others might not consider this to be a bad deal overall.