I rountinely use LLMs for cleaning up prose in various reports. As a proofreader and editor, it's the best tool I've ever found.

For creating reports from scratch, you have to be careful. It's not perfect, but it will get you 85% to 95% of the way there on a first cut once you feed it the data. It's no replacement for a human, but it does save a lot of time.

I also use it for email. When I have an important email to send out that must be "perfect", I'll run my draft through ChatGPT and ask it for a review, and to show me what it changed, and why. More than once, it has caught a missing word or a clumsy phrase.

So yes, LLMs are not a gimmick, and they do increase productivity if used correctly.

This year, enrollment grew by only 0.2 percent nationally, and at many programs, it appears to already be in decline, according to interviews with professors and department chairs.

At my university, we've seen an 8% decline in CS enrollment since Fall 2022. It will probably be even worse when the Fall 2025 numbers are tallied. On the plus side, students are rapidly moving back into other engineering majors again, with several seeing 20% to 60% growth over that time.

Unfortunately, our adminstration decided back in 2021 to put all their eggs into the CS basket (i.e. faculty hires, budget allocations), just before the CS bubble began to burst. Other engineering programs are now desperate for money as their enrollments soar, but I expect that it make take a few more years before the message percolates up to the Chancellor.

Heaven help all colleges where non-STEM administrators decide to implement their visions about the future of STEM.

Yup. They are every bit as culpable for climate change as the fossil fuel companies. Now they complain it will take too long. Well it has taken a long time for us to get to where we have 400 reactors and 12,000 thermal coal plants in the world, so it is indeed going to take a long time to fix. Best start now rather than whining about how long it will take, that's the reason we are where we are in the first place.

Or to paraphrase a well-known saying: "The best time to build a nuclear power plant was twenty years ago. The second-best time is today."

The rise in the use of signal jammers is due in no small part to the popularity of WiFi-connected security cameras. For example, signal jammers are very popular tools among professional burglary rings. They've hit multiple homes in the pricier areas of the county I live in.

Companies like Ring have been very successful in convincing consumers that "Ring = home security", when in fact Ring cameras are a terrible choice due to the fact that they must have a WiFi connection in order to operate. The burglars scout out pricey homes in neighborhoods where no one ever talks to their neighbors, and houses are isolated by distance, foliage, and fences. And it just so happens that the owners installed Ring cameras.

So the burglars drive up, activate their jammers to knock out both the Ring cameras and the cell connection for any burglar alarm, then clean the place out. Even if an alarm goes off, chances are that no one will hear it, or call the police if they do.

If you're truly serious about video security, you need hardwired cameras with onsite storage and battery backup. You might still get robbed, but at least you'll have something to show the police.

The interesting part is the human psychology behind it, i.e. what causes (allegedly) intelligent people to perform a necessary test, and then simply ignore the results of that test when the results aren't what they had hoped for?

That type of behavior (ignoring inconvenient facts that fail to fit your desires or opinions) is so common that it's practically the norm for humanity. Someone who can look at the facts and accept a conclusion that will cost them money or social standing is a very rare bird indeed.

Rush only killed four people besides himself. Politicians with similar mindsets have been responsible for the deaths of millions. In that sense the world is lucky that Rush's influence only extended to OceanGate.

You could replace "mosquitoes" in that headline with polio, smallpox, measles, AIDS, malaria, or any one of a thousand different pathogens. No one is agonizing over eradicating them.

If a particular species of mosquito is a vector for a deadly disease, eliminating that one species provides a net benefit. You don't have to kill all mosquitoes, just the species that are truly dangerous. So yes, do it.

So are you advocating for the repeal of all the bill off rights, or just this one?

No, I'm pointing out that aerial photography of private property is absolutely permitted (and legal) under specific circumstances. I can go to Google Maps right now and see which of my neighbors have swimming pools and decks. So where was the ACLU back when Google started doing this? Clearly the 4th Amendment is not preventing Google, Apple, or anyone else from taking aerial photos.

The gist of these complaints is "I spotted a drone near my property, so the government must be spying on me!" There is zero proof from the news article, or from the ACLU brief, that the county has captured any information that could not have been captured from a helicopter or aircraft. The drone is just a lot less expensive.

On top of that, the people who are complaining are doing so because their own local govenment cited them for codes or zoning violations. It is the job of your local government to enforce those regulations. Just because you built that unpermitted hot tub behind that 8 foot fence and got caught doesn't make you a victim of illegal surveillance.

You can take pictures from the street in front of your house but not from the air? Hard time understanding this.

For that matter you can also take photos from an airplane, helicopter, or satellite. So where's the outrage there?

There is no indication that anyone's "privacy" was violated beyond their desire to keep their code violations hidden from regulators. You can see the image in the SFGate article that the woman is suing over. You'd be hard pressed to spot a human figure in that photo, but you can definitely see the hot tub she built.

That just means that most of the writing you do is useless or of negative worth.

No kidding! You are absolutely correct. These reports that I must generate as part of my job are the type of meaningless verbiage that middle-manager types with too much time on their hands demand on a regular basis. They're a waste of electrons.

Does anyone actually read them in detail? Of course not, but that doesn't change the fact that someone wants to see them.

The modern world is full of pointless documentation and record-keeping. An LLM can save you countless hours creating it, if you use it intelligently and double-check what it produces.

Exactly what do LLMs "produce"?

If you expect them to act as finders of fact, or to think creatively, you'll be disappointed. But if used properly, they dramatically increase productivity. LLMs are the best writing assistants you could ever hope to find.

Yesterday I uploaded 22 separate status reports into ChatGPT, then asked it to summarize the findings and plot the statistics. It did those things flawlessly, and in a matter of seconds. It would have taken me hours to copy & paste while compiling the numbers. ChapGPT provided a report in Word format along with tabulated results in less than a minute.

So yeah, a lot more useful than Bitcoin.

Are they getting trolled here; or do the fundamentally misunderstand why people find them creepy? This is basically "Zuckerberg thinks that AI friends are the solution to loneliness" the movie"

As much as I hate to give credit to one of the world's leading techno-sociopaths, Zuckerberg isn't entirely wrong. My wife is a social worker who works with lots of shut-in patients, and in many cases even an AI companion would be better than the complete lack of human companionship and interfaction that many of them have. There are a great many depressed and lonely people out there, particularly since the pandemic.

Always having someone to talk to who cares about them and never gets bored or frustrated talking to them would be a boon to some patients, if it was done properly. On the other hand, I would never trust any AI companion that Zuckerberg or Meta came up with.

I've long wanted a bright array of IR lamps to put around my license plate holder to blow out traffic cameras....

Too late for that. Camera manufacturers are transitioning to white light illumination with IR filters due to the prevalence of 3M printed plates which have terrible IR reflectivity. They'll even read paper plates at night.

People speculating about how we're going to round up or round down without pennies are missing the point. Since COVID, I rarely carry cash, nor do the people I know and work with. It doesn't matter if pennies aren't minted anymore, because hardly anyone is carrying or using them. Practically everything is cashless.

The one-cent denomination will survive just fine. If I buy something that costs $21.62, then that's what I'll be charged on my credit card. The round up / round down issue only matters if I pay for something in cash, in which case the retailer will decide how to handle it.

Thirty years ago, there was a fierce debate on abolishing the penny. But today? No one cares, because for most of us it no longer matters.

I do not see current "AI" technology as being much of a threat to STEM fields because ultimately our fields concentrate on understanding concepts and using that knowledge to do things that nobody has done before or apply those concepts to new situations.

I wouldn't be so sure of that, given the impact that AI has already had on computer science. I think it's already impacting CS enrollment.

Current "AI" does not understand concepts at all, it is a predictive text engine who can parrot its training data to make it sound like it understands concepts but really does not and, because it relies on its training is not great at coming up with completely new things or dealing with new situations.

I can't help but wonder if human intelligence isn't pretty much the same thing - a predictive pattern-matching engine running on a biological substrate. Let's just say that I suspect that we may be closer to AGI than many people believe. We shall see.

One might also argue that we're but years or months away from labeling a college graduate as nothing more than someone who hopes their $60K+ investment in a Bachelors of Artful Scamming is going to be valued outside of a college campus marketing bubble.

As an engineering professor, my prediction is that the economic value of a liberal arts degree is going to fall nearly to zero by 2033 at the latest.

My reasoning: the kids who started using ChatGPT to cheat on everything are currently in 8th or 9th grade. Give them 7 or 8 more years, and they'll be graduating from top tier colleges with reading, writing, and reasoning skills that have barely advanced past elementary school. Then companies are going to interview (or hire) them, and the jig will be up, so to speak.

It will be an apocalypse for liberal arts education everywhere. Ivy League universities will find that no one wants to hire their functionally illiterate graduates. And in fact, it will make more sense to just use an AI that by 2033 will be far more capable than most college graduates.

The end isn't quite as close for engineering. No amount of ChatGPT cheating will help a student pass an exam in (for example) circuit analysis right now. But in the long run, no degree program is safe.