Closi

There is a trade off with simplicity though - usually it requires being highly opinionated about which software features make the cut. Users or sales teams often want more features, but if you included every one the app would become a mess.

But there is a possible world where you can have both - every 'feature' your users would ever want without overwhelming complexity or steep learning curves, but with the possible downside/cost of reducing consistency.

It's still better than anything else and the de-facto standard so you need it.

Clients will send you their PowerPoint template and want you to use it. They will send you their complex spreadsheets riddled with VBA macros and you will need to fix them. They will invite you to a Teams site because that's where their project updates go. I just don't see how you can avoid it as a consulting company!

For things like Excel - We can say it's 'bad' but I've not seen anything do the job it does better. And besides, even if it was bad, it doesn't matter - as a consultant you need to use it because your clients probably want your workings as part of the deliverables, and if it's on Google Sheets they often won't want that.

You can run a logistics consulting business without windows, but you will struggle without Excel and PowerPoint, and 365 with SharePoint is basically needed for collaboration in any consulting business.

Im also a logistics consultant… try to parse a multi-million line orderlines extract in Google Sheets compared to excel.

I’m also on Mac but to be honest it’s a challenge - there are still enough industry specific tools that are windows only so I have to run a parallels VM to get by.

> we reach to a safe and eventually perfect system by deploying it in public and iterating over it until optimal convergence dictated by users in a free market

Possibly... but also a lot of the foundational AI advancements were actually done in skunkworks-like environments and with pure research rather than iterating in front of the public.

It's not 100% clear to me if the ultimate path to the end is iteration or something completely new.

Terrence Tao seems to think it has it's use in finding solutions for maths problemms:

https://mathstodon.xyz/@tao/115591487350860999

I don't know enough about maths to know if this classifies as 'improving on existing results', but at least it was a good enough for Terrence Tao to use it for ideas.

I'm just going on the fact that most energy experts say a diversified mix is what you usually want - although my lay assumption would be that if you have a portion of your energy need covered by a stable base load, you need less batteries per wind turbine etc and it lowers overall energy risk.

e.g. lets pick two overly-simplified hypothetical scenarios where you can have a 50/50 nuclear/wind mix, or 100% wind mix.

And then lets say in a shortage scenario you can: - Suppress demand / demand shift (say you can reduce/shift demand 10% hypothetically)

- Import energy, up to 20% of your total requirement

In this simple scenario, in a 100% wind scenario, you would need to cover 60% of your energy need with batteries (every 1% of wind needs 0.6% of batteries), but in the 50/50 mix scenario the 50% of wind only needs to cover 10% with batteries (i.e. every 1% of wind needs 0.2% of batteries)

In reality there will be other things, like you might also have burst-gas power etc.

I'm not saying the numbers above are correct - just trying to show in theory that a mix / base load can still help reduce the level of batteries/storage required and make the energy mix more economic.

> You can see on the map at the bottom of this page that almost all the batteries are in areas that already have high amounts of renewables:

It could be - but the battery investments map also align with the map below which shows that these states (Texas & California) are also states suffering from blackouts.

https://worldpopulationreview.com/state-rankings/power-outag...

So while this could mean that storage is cheap, it could also mean 'Texas's mix and grid is unstable, particularly as it's not connected to the national grid, and this has opened the opportunity to profit from higher levels battery arbitrage that doesn't exist in a better balanced grid'

> Net result renewables currently save you money until ~80% annual electricity supply. At which point adding more batteries and generation to cover overnight demand is cheaper than adding nuclear to the mix.

Assume you mean more expensive than nuclear in the second point?

Agree with your point although it's about wind in the uk rather than solar, and about being able to last a few weeks if there is calm weather rather than a day without sun, which is when having a nuclear baseload makes sense.

> This is shortsighted because [...] electricity renewables are cheaper than nuclear

This is an oversimplification - Renewables are cheaper than nuclear, but they are also less reliable than nuclear in the sense that when the wind stops blowing, power stops being generated. Also if you include the cost of energy storage to survive a week or two without substantial wind, suddenly it's not the cheaper anymore.

A mixed nuclear + renewables grid would reduce the total cost because nuclear can provide a stable base load which isn't affected by seasonality. Modern plants can also ramp up/down to some extent to balance the overall system.

That's why you need an energy mix rather than just putting all your eggs in a single source.

To be honest this is the sort of thing Nano Bannana is weak at in my experience. It's absolutely amazing - but doesn't understand left/right/up/down/shrink this/move this/rotate this etc.

See below to demonstrate this weakness with the same prompts as the article see the link below, which demonstrates that it is a model weakness and not just a language ambiguity:

https://gemini.google.com/share/a024d11786fc

> AI will never know the why behind the architecture, it will only take it at face value.

There is no reason to believe that at some point in the future AI will know the business context for apps when they are vibecoded (because the prompts should contain the business context).

Amazon?

https://www.seattletimes.com/business/amazon-pays-off-its-hi...

Also 5% would be a ridiculously low rate for this sort of corporate finance. You would expect more like 8-12% I think?

Plus the post seems to only include 1 year of interest.

Unless we know the terms, I don't think we can necessarily calculate EV from JP Morgan's perspective. I would say that they aren't usually carelessly giving away money though... They probably have terms where they can get out early if OpenAI's position weakens etc.

I don't know if I agree in this instance. While I agree that Black people completely have a shared identity and culture - the article is clearly talking about skin colour and it's doing a comparison between how AI represents two skin tones, so I would assume that by your definition it should use lowercase in both cases.

If it's comparing a culture vs a 'non-culture' then that doesn't sound like for like.

That depends - scrapers are currently annoying and temperamental and you have to maintain them. Also the idea of allowing some random person from Fiverr to write some code you are going to have running in your infra that has access to your webshop, ERP and the open internet isn't usually that palatable to most IT teams.

A simple and reliable API? That's different.

The counterpoint is 'Why hand your competitors data on a silver plate'?

Sure you might be willing to build the bot to scrape it... but some other competitors won't go to this effort so it still means a bit of information asymmetry and stops some of your competitors poaching customers / employing various marketing tactics to exploit short term shortages or pricing charges etc.

While I agree it was ultimately flawed, I think it's likely that the core team at Humane genuinely thought this was the future of computing, and clearly put in a lot of effort.

... And maybe something like this is, it was probably just too early.

I know, that’s not what I was saying :)

> Chinese people aren't stupid, they have been spending a fortune on automating as much of their manufacturing as possible!

Slight nuance - they have spent a [reasonable amount of money] automating production.

The trick to automating something that ‘isn’t a car’ is often to put in small bits of low-cost and flexible automation that can be moved around and repurposed. IMO this is often what we are bad at in the west - companies can/do setup massive automated sites at huge expense, but there aren’t the skills/infrastructure to do this at the lower end of production (eg if you want to deploy one AMR in the west the AMR companies don’t want to talk to you, and there isn’t really an easy way to get one yourself without talking to an integrator which will charge tens of thousands which will wipe out the benefit, and we don’t have the skills within most small production companies to get a small robot arm/AMR working without external integrators - but a one-AMR deployment might be a more common scenario in China).

I think they are referring to:

> Durham University improved by 30 places year-on-year in its students’ evaluation of teaching quality, which was the main driver in securing its third place in the overall university league table

Which isn't quite the same as 30 places in ranking as OP suggests, however I agree with their point that moving 30 places on that metric could be fairly suspicious.

For example - when I was at university in the UK we got a speech telling us basically that we were going to get sent a survey from the times, and the higher we ranked the university, the higher the universities ranking would be, and that would make our degree more valuable. If the main reason they jumped from 7th to 3rd could be a metric that is potentially 'influence-able' by the university, it could be more of a change in comms-strategy than actual university quality.

Very true, although even at 2 or 3 drivers it’s still more than months.

> maybe they are at around $200k/vehicle? That would be pretty easy to pay for a with a few months of rides.

The 'few months' bit doesn't seem quite right - the cost to get a human to drive a car would maybe be $50k per year, so i'm not sure how a $200k vehicle can pay itself off in a few months vs a $50k car + $50k per year driver.

I'm aware that the cost / ride is higher for Waymo, but it doesn't sound like that would be enough to cover the extra $200k and not certain that scales to other geographies outside of SF.

I mean to pay off a $200k vehicle in a few months you would need each car to be clearing $3k a day in revenue or something like that.

It's probably a few years per car if they are at $200k. If they are 'in the black' or not will probably depend more about their accounting rules (i.e. depreciation) more than anything else.

IMO the benefit of a local LLM on a smartphone isn't necessarily compute power/speed - it's reliability without a reliance on connectivity, it can offer privacy guarantees, and assuming the silicon cost is marginal, could mean you can offer permanent LLM capabilities without needing to offer some sort of cloud subscription.

I'm not entirely sure the analogy is fair - Amazon for example was 'ridiculed' for being hugely unprofitable for the first decade, but had underlying profitability if you removed capex.

As a counterpoint, if OpenAI were actually profitable at this early stage that could be a bad financial decision - it might mean that they aren't investing enough in what is an incredibly fierce and capital-intensive market.

Exactly this - people that understood the field well enough to add new knowledge to it has to be a pretty decent signal for a research-level engineer.

At the research level it’s not just about being smart enough, or being a good programmer, or even completely understanding the field - it’s also about having an intuitive understanding of the field where you can self pursue research directions that are novel enough and yield results. Hard to prove that without having done it before.

> All the other things he mentioned didn't rely on breakthroughs, LLMs really do seem to have reached a plateau and need a breakthrough to push along to the next step.

I think a lot of them relied on gradual improvement and lots of 'mini-breakthroughs' rather than one single breakthrough that changes everything. These mini-breakthroughs took decades to realise themselves properly in almost every example on the list too, not just a couple of years.

My personal gut feel is that even if the core technology plateau's, there's still lots of iterative improvement to go after on the productisation/commercialisation of the existing technology (e.g. improving tooling/ui/applying it to solving real problems/productising current research etc).

In electric car terms - we are still at the stage where Tesla is shoving batteries in a lotus elise, rather than releasing the model 3. We might have the lithium polymer batteries, but there's still lots of work to do to pull it into the final product.

(Having said this - I don't think the technology has plateau'd - I think we are just looking at it across a very narrow time span. If in 1979 you said that computers had plateau'd in 1979 because there hadn't been much progress in the last 12 months they would have been very wrong - breakthrough's sometimes take longer as technology matures, but that doesn't mean that the technology two decades from now won't be substantially different.

I guess different small models will have different points/goals, but you can still have a small model with lots of training effort or a large model with little training effort.

I think the point of most (frontier) small models is usually to provide the best answer possible given small inference resources, rather than to reduce training time.

This is more of a toy model, so fun and an interesting project but it doesn't necessarily tell us what the art of the possible is for small models.

I don’t think we should use an AI trained in 5 minutes on a laptop to infer what small models are capable of…

Sure they still have massive problems with hallucination, but this article doesn’t give us any more insight into that I don’t think!

I assume you don’t use these models frequently, because there is a staggering difference in response quality from frontier LLMs compared to GPT 3.

Go open the OpenAI API playground and give GPT3 and GPT5 the same prompt to make a reasonably basic game in JavaScript to your specification and watch GPT 3 struggle and GPT 5 one-shot it.

> Suppose it can't. How will you know?

By finding and testing problems that o3 can't do on Deep Think, and also testing the reverse? Or by large benchmarks comparing a whole suite of questions with known answers.

Problems that both get correct will be easy to find and don't say much about comparative performance. That's why some of the benchmarks listed in the article (e.g. Humanity's Last Exam / AIME 2025) are potentially more insightful than one person's report on testing one question (which they don't provide) where both models replied with the same answer.