Gotta agree. I can say with respect to medical insurance, Bill Gates is pretty much dead wrong. AI can't make medical claims easy if the people who approve the claims change the rules with no notice. Which they will, because claims management is a constant push and pull over who keeps the premium dollar not a high school history paper.

  I recently heard a VP at Geisinger say that their efforts to use current "AI" in charting had also not yielded meaningful efficiency gains.
That's because the humans in the systems still are and must be responsible. If you know the RACI scheme, consider this: Responsible = humans, Accountable = humans, Consulted = LLM + humans, Informed = Humans.

The basic problem is the same for doctors as for drivers. The human is accountable, but to understand and take action the human must learn. However good the LLM is for a human to learn requires thought, effort, and time. Basically when a doctor lets "AI" write the clinical note, s/he then has to supervise by reading the note and fixing it. That takes more work than just writing the note.

And unlike LLMs, humans can actually reason and intuit,so when faced with unfamiliar inputs they can in fact exercise judgment and common sense. LLMs cannot. They also cannot revise their prior training in realtime as they learn which facts to believe and which not to believe.
And when the shit hits the fan, some human needs to be able to explain, reason, and take responsibility. Doctors cannot do that without charting, drivers can't do it without paying attention.

Yes, LLMs can help gather and synopsize information. They are likely to be used for this, though they would sure be more useful if one could understand and constrain the on which they were trained. E.g. for a doctor, an LLM trained mostly or entirely on medical literature is vastly more useful than one trained on the internets. And even within that set, one would need to weed out publications that proved to be simply wrong or even fabricated (Lancet article on autism anyone?).

LLMs could well save us software folks a lot of frustrating debugging time - in fact I think that's one of the best use cases around for them.

But take over tasks for which humans have responsibility? Not for a while.

It's an interesting idea - I've always thought that some tax code structures to incentive lowering rents would be a good thing.

However, I'm not sure that will work for this particular problem because there may not be enough demand for office space to keep these buildings occupied no matter what the rent is.

I also have read stuff about provisos in the original financing of these buildings that establishes minimum rent. Similar to startup valuations, I think there's a strong incentive for both lenders and property owners to wait and hope that things get better. After all, write-downs of the value of the property mean pain for all parties.

If that is true, then I do think it would be worth exploring regulatory and tax changes that encourage the property owners and lenders to do debt write-downs faster rather than waiting until they have no alternative.

All that said, I think the Post editorial board is kind of hyperventilating on this one. It is probably true that these downtowns will get somewhat rougher if there is persistent vacancy. But as someone who grew up in DC in the 80s, we've already had that and we survived. And a reduction in the fanciness might not be all bad news if it allows ordinary people a little space to exist and flourish.

It's also far from clear that the immediate action they want would even be effective. Let's say that some city plan manages to get 30% more buildings to go residential than would otherwise do so (a pretty optimistic thing given the timeline for permitting, financing, and even assessing residential demand). That still puts the new residents in the middle of a lot of vacant office space, at high cost to taxpayers and crowding out other spending needs.

Plus, there is still a lot of residential building going on in other neighborhoods, so the downtown properties would be competing for residents with those neighborhoods and buildings.

If any money was going to be spent, I'd much rather see it go toward incentives for small businesses and others to operate in the existing structures at lower cost.

I'd say the problem there is one of lack of depth. Of course the President of a Methodologist college doesn't know how to discern whether a portrait of Mohammed is offensive - that would take a great depth of experience with Islam, some scholarship (both religious and secular), and even then you'd probably still arrive at an answer that is unsatisfactory to many. But I don't think you can just write off the concern of the student either (AFAIK it is true that many practitioners of Islam consider portraiture of Mohammed to be blasphemous).

I honestly am not sure that there's a clear answer there, other than that university (and other) administrators need to find a way in which they are responsive to real concerns while not being less deeply involved in what individuals care to say or do.

Right now it seems that many actions are reactive and driven by fear of negative press from a host of stakeholders.

Yeah, to be honest I think story is more indicative of the fact that "Speech First" succeeded in hooking a reporter on some bs than any real indication of the overall environment at Stanford or any other college.

Pretty much 0 real data under discussion here.

This issue is so frustrating.
One the one hand I do think there is a prevailing culture in which people are uncomfortable saying or doing things because they fear being judged for whatever people perceive about it that, and how it may get interpreted without offering the benefit of the doubt to the speaker.

One the other hand these Speech First guys bundle supposed free speech with opposition to any acknowledgement or attempt to remedy actual, real racism.
https://speechfirst.org/share-...
Any time you see "DEI" and "indoctrination", you can pretty much rule out any intent to actually foster an environment where real issues get discussed openly. It's pretty much just reactionary bullshit that insists that racism is suddenly "over" and we can no longer talk about it.

You can in fact have DEI and free speech. You just have to be ready to do some hard work - acknowledge that racism is real, but also that people's specific word choice does not always fully reveal their motivations or the extent of their perspective. And some people are going to have to reflect on how they came to their station in life - almost always some mix of luck, social circumstance and hard work - without the comforting assumption that things were all fair and equal along the way.

It seems that now there is neither, and that is bad for everybody.

The problem is that lighting does much less to deter crime than people think, and most "security" lighting is done poorly and not as a substitute for well-planned ways to make a built environment safer.

Even studies that seem to show positive correlation between more lighting and crime reduction are highly situational:
https://www.cnn.com/2022/04/28...

Other studies have shown increases in crime related to lighting (because people can more easily congregate and do illicit things).
https://popcenter.asu.edu/cont...

And in any event, the best answer is rarely what people do. They go get poorly targeted lights that generate a lot of glare and deep shadows, which are just as likely to promote crime as to reduce it.

So now there's light pollution, wasted power, sunk costs on ineffective infrastructure, ill effects on human health from disrupting our physiological response to darkness and light, but relatively limited effect in abating crime.

Another problem is that trees have to be watered and tended in order to survive. A depressingly large proportion of all the trees planted in a year don't make it more than a few months before dying.

Yeah, the other problem that gets no mention is about surface area. Because atmospheric CO2 is diffuse, you need to make contact with a lot of air in order to remove a lot of CO2. That's actually quite hard to do with any sort of mineralization approach. The trays these guys are showing in their videos would need to be absolutely everywhere in order to get anywhere near the kinds of numbers that they're throwing around.

As someone else pointed out, the thing that can most easily cover wide surface areas is plants. So the single fastest path to get carbon out - well behind reducing emissions of course - is to get plants to capture and sequester more CO2.

One interesting angle for this limestone approach is the possibility of de-acidifying the ocean. If they can control the dissolution or recapture the limestone after dissolution, then this could be a valuable complement to other methods of removing carbon from the ocean.

I don't really know what I think about Tesla's build quality, and I'm definitely not in a position to argue the facts you are citing so I won't. But I do think it's worth pointing out that your argument about people's satisfaction overlooks so important things about a measure like satisfaction and perception of value that do apply in this case.
As you note, the satisfaction statistics reliably indicate that people *feel* they got their money's worth, and that perception is the fundamental measure of value. However, what they feel they bought may include:
- An electric car in a moment when such a thing was futuristic and "impossible"
- A sense of being environmentally responsible
- A feeling of style or status from having something unique

These sorts of perceptions are a essential part of the value of any car. However, some of those perceptions may change as the market shifts around Tesla.

For my own part, the couple times I have driven Teslas I've enjoyed it - fantastic acceleration and good road feel. I am excited about EVs and want to own one (currently I have a plugin hybrid Pacifica and I love it). But I don't want to own a Tesla - the interior style, the dashboard UX, and the price point are not for me. And I'm honestly excited to see the EV market fill out.

All that said, everything I've seen and read says that Tesla still maintains a significant lead over most other players in range/batter energy density/efficiency, excepting perhaps Lucid and a Chinese manufacturer or two. Even Volkwagen's CEO has basically come out and said so. So I'd say I'm with you that the analysts claiming Tesla will be overtaken next year are probably blowing smoke.

Not sure I agree with that. Smart people want to build stuff, and the strategy of pursuing long-term nebulous goals by hiring lots of people does not readily equate to building stuff.
I think he's probably right that Facebook could actually speed up its development by not pursuing such a big broad strategy. And setting solid mid-term benchmarks of success is also a good strategy - it means that people can understand their progress and adjust.
OTOH, *how* they execute "streamline" matters as much or more than exactly how many people they cut. Once an organization gets bloated it's often pretty hard to figure out what is muscle and what is fat. I also think they have to address their damaged reputation and find something exciting and meaningful to attract good people. People who want to build stuff are not going to be excited about a company with a reputation for abusing users' data without being able to point to something cool that those people can say they built.
I honestly would not be sad to see MetaFace drift into oblivion but I do think this is actually sound advice.

That's been true thus far, with the added fact that they have built a significant business model around cross-selling services based on their device/app store monopoly.

As a result of that keep-it-the-company business model, Apple has had almost no scrutiny of the data they do collect. I'd also bet that they have very little internal auditing of that data, and a culture of sharing it freely for cross-selling purposes.

We really have no idea what Apple does with iMessage data or metadata, while we do know that Facebook keeps and monetizes WhatsApp metadata and has made noises about breaking end-to-end encryption so it can monetize content in WhatsApp. This makes for a pretty poor set of choices for consumers.

I'd love to see federal legislation requiring Apple/MetaFace/Google and others to state specifically what they going to do with consumers' messaging data (at a minimum), including a do not monetize option.

Interesting. I have a related take. Technology takes off when it meets a need that people have (whether they know it yet or not).

As you note, tech companies often fail when they try some sort of extrapolation from their current business to future needs. This is compounded by the fact that these billionaires rarely fail to be seduced by the flattery that they are somehow special visionaries rather than smart people at the right place at the right time (Exhibit A: Musk, Elon). And further compounded by the fact that the new businesses may conflict with or cannibalize their old business.

In the case of VR, I don't think anyone has discovered the need, latent or not that VR meets. The need Facebook initially met (keep in touch with people you might not otherwise be in touch with) was an actual need that people had. If nothing else, Facebook created a directory app where you could find your friends and acquaintances. But now that those capabilities exist and have matured, and Facebook's built a business on monetizing attention and your social graph, they are casting around for a way to extrapolate their business to something else without regard to what sort of unmet needs people have. In addition, the world has since seen a lot of downsides of having technology mediating social interactions, and even worse when the technology is owned by investors who demand incessant growth in revenue and profit by monetizing the user even more, and there's justifiable public skepticism about Big Tech.

The iPhone was far from the first smartphone - what was a game-changer was the App Store and streamlined delivery of apps and content. That allowed people to easily get apps and use their smartphones for things they knew they wanted (listening to music, videos, etc) and things they didn't know they wanted (getting taxis and booking stays at other people's properties on their phone).

I don't see anything in VR that unlocks unmet needs (latent or known) any time in the near future. Gaming will probably progress toward VR but along game-specific use cases rather than some general social world. The only need that's close is remote work, and I don't see Facebook looking seriously at how to make the interactions in remote work better. I will not be sad to see this Metaverse wither away.

Couldn't agree more - this is Signal putting its goals ahead of its users'. My goal as a user is to get as much privacy as I can while communicating with the people I care about. To ask me to put someone's privacy agenda ahead of communicating with my people is absurd. I will almost certainly drop Signal when this happens.

They are abandoning the mass market in favor of remaining a niche product. From an organizational/business perspective this may make sense for them, but in terms of their mission-oriented stuff to improve privacy for everyone it's a big step back.

From a landscape perspective, another reason this will not work is a single app: iMessage

iMessage is central to the iPhone/Mac experience and most iPhone users simply will not drop it, nor do they have any desire to do so. Apple knows that iMessage is pivotal to its userbase and guards it zealously. If Signal doesn't interoperate with iMessage, anyone who communicates with someone on iMessage will have to drop Signal.

Since my wife and daughter have iPhones and rely on iMessage for various features (e.g. FaceTime), I will have to use something SMS-compatible to talk to them. While some friends are on Signal, it's too much trouble for me to deal with them out of band. I might keep the app for calling but otherwise I'll be out.
Pity, I liked the privacy when I had it.

For context, Obamacare enrolls https://www.kff.org/other/stat...

1- Employer-sponsored (158M/ 49.%)
2- Medicaid (63M/ 19.%)
3 - Medicare (45.2M /14.2%
4 - Uninsured 29.3M (/9.2%)
5 - ACA/Obamacare (18.7M/ 5.9%)
6 - Military (4.3M/1.4%)

The bulk of cost problem comes from employer-sponsored insurance, which typically pays 2-3x the prices that Medicare pays. Additionally, ~50% of Medicare beneficiaries are actually covered by private plans, which are notorious for overbilling:
https://www.nytimes.com/2022/1...

Note as well that the CBO's most recent report on medical costs also indicates that price regulation is likely needed in order to contain costs:
https://www.cbo.gov/publicatio...

On the other side of the coin, most countries wait for the US to prove out new medical technologies. US companies lead the development of new technologies across almost all disease categories. Essentially no other country evaluates clinical efficacy at the level the US FDA does. Our insurers routinely pay for technologies that others pay for many years later. This means that in terms of getting access to the latest technology and the physicians who have used it, the US is the place to be.

We pay a very dear price for that position, but it's not clear that our public wants to relinquish the access to new technology.

The problem is that any change is basically political death - people are very afraid of losing the coverage they have even as costs spiral out of control.

Politically, the only way I can see to gain control is Medicaid for All (not Medicare for All). That is because Medicaid is part-federal, part state - roughly 50/50 split of costs. Having Medicaid for All means that any provider or drug/device manufacturer seeking high prices has to knock on at least 50 state doors to get coverage. Conversely, with Medicare there is only the federal government in the way of charging whatever they want (really there are several MACs deal with coverage determinations but that's a little weedy for this conversation and they all basically follow CMS anyway). Given the exorbitant prices charged in the other major monopsony - defense - Medicare for All would likely be a blank to providers/manufacturers.

Great comment. Obv I'm late to this discussion, but I think you're raising something worth considering more deeply when this topic resurfaces.

The logic of a meaningful program is usually quite complex, and no matter which form of representing the logic you choose there is irreducible effort and complexity in understanding and working with that logic. Code is often the most concise and usable representation of that logic. At my org as we've been doing data pipeline work, I've studiously avoided the flow chart method because it tends to obscure many of the key questions. For example, if you have a decision point in a flow chart, often there is validation logic and a wide range of possible exception flows. Trying to put these on a flowchart makes a mess and reduces clarity rather than adding it. Even worse, it allows people to avoid organizing exceptions into classes, which is often critical for making good code and for refactoring entire processes. The code approach actually drives much more clarity.

As you get better at engineering (SW and other kinds), you develop skills in problem definition, problem decomposition, implementation evaluation, architecture, evaluation/prognostication of what will happen over the project lifecycle, etc. In fact, one of my benchmarks of a great engineer is the degree to which they can do all of these before writing code. Low code projects without these skills, will, as others pointed out, be poorly organized code. My claim is they will also be poorly organized projects, unless a good engineer makes those things happen.

Where I see utility in low-code solutions is more in prototyping and in backstopping good engineering. If you build a low-code solution that helps you develop a better understanding of the problem and iterate past bad implementations more quickly that's great. And if you use AI code generators to create reference implementations for comparison that might be useful too. I just really doubt that it's going to be a substitute for real engineering skills any time soon.