Re: Nuke codes?
I heard the British codes are only 3 digits, because they are big Ian Flemming fans... 007 ftw.
7078 publicly visible posts • joined 27 Nov 2009
I agree, although, on the other hand, I'd rather a rogue AI agent under the auspices of training from somebody like Anthropic or OpenAI found the security flaws in my defences, rather than an AI working on behalf of a ransomware gang.
Ideally, neither should be attacking my network, and OpenAI and Anthropic should at least be censured, if not fined, for the attacks - gaining access to a computer not belonging to you is illegal, whether you do it, or an AI you control does it...
(I'll politely assume for now that this is an inadvertent misfeature of MacOS 27 still being in beta status, but if this is still the case come the actual production release…)
Exactly. This is a beta and betas break things, that is why they are only rolled out to a limited number of testers, who agree to give feedback about bugs they discover. This has been raised as a bug, and it shouldn't affect any production machines, because, you know, you never put Beta software on a production machine that you rely on, you always put it on a test machine... Right? ... Right?
Totally, except this isn't an update, it is a limited availability Beta to find out what the changes broke... Those who signed up agreed to use the beta and that it would be faulty and unstable. If this persists into the actual update/upgrade, that is a whole different kettle of fish.
If it were actually an update, I would agree. This is the first developer beta and thus beta software, which is buggy. This sounds more like a simple case of Hanlon's Razor: never attribute to malice that which is adequately explained by incompetence/stupidity."
Betas often cause things to break, that is the whole point of a beta, to get it out from developers to testers, to find out what the developers didn't think about and that broke things that the developers didn't consider. This isn't a final release going out to a large audience, this is a limited beta going out to people who signed up to test the new system and provide feedback on what doesn't work.
The image server is on a 1gbps connection, and we generally have half a dozen PCs on the workbench being imaged. We aren't waiting for them to complete, we are usually off doing support or other tasks so they will probably sit there for half a day, before we get around to the next steps.
Adding to the domain takes about 1 minute, including booting and renaming. I've never had it fail. Logging into the M365 account is the user's job and, again, rarely fails. M365 downtime? There was slow email this week, but not had an outage during working ours for ages. I used to joke about Microsoft 365, 360, 354, but, honestly, the reliability these days is pretty good.
All the software we use is very stable. apart from the development releases of connectors on our test ERP server, which talks to the production management server. I haven't had reports of crashing software for over year, other than the ERP system, where the supplier delivered a faulty update (JAVA, so cross platform, it would have affected Linux, UNIX or macOS just as badly). As for Windows stability, our users are quick to moan, if a PC crashes. We have had maybe 5 reports in the last 12 months, 2 were RAM problems (older PCs still on 4GB)†, one a knackered SSD and 2 dead batteries on a laptop.
Reprovisioning a Windows PC is an unattended process (after the inital PXE-Boot). It takes between 30 and 90 minutes to complete - depending on the PC and the required image. After that, it needs to be added to the domain, then the user needs to log on with their M365 account and the rest of the software gets pushed and all policies applied... It works very nicely.
The software we use runs fine, no compatibility of stability problems.
I use macOS and Linux at home, and I have the only Mac at work, but I have to configure and support Windows PCs and servers, plus a couple of Linux servers, although a lot of time is spent working on Windows Terminal Server.
The wrong analogy. You pay to run Excel on your local computer, so your local computer and local electricity bill is taking care of that load. With AIaaS, the load is running on somebody else's computers and they cost a fortune and cost a fortune to run. At most, with M365, you are paying to have the files stored in Microsoft's cloud, and you get a fixed sized quota that you can store, if you want more, you have to pay more. But it is a known quantity, you aren't paying Microsoft for 1TB of storage, but you are actually storing 30-40TB a month...
The other thing is, Microsoft sells 1TB to each customer, but 99% of them only use a few GB, so they don't need to have 100% coverage for storage on day 1, they can build it up and the low usage users are subsidising the high usage customers. When it gets to a certain size, they have to raise prices, because they need to invest in more storage and more quantifiable compute power for the web services.
But, on the AI front, even the most lightweight user is using more compute resources than their subscription pays for, so there is no unused quota that can be put to high usage users to offset them. EVERY user is losing them money. Now people have become used to using it, they are now being asked to pay for what they use...
Exactly, these companies have been racing to the bottom by offering their services at a loss, trying to gain customers, now they need to actually cover their costs and people complain... Go figure.
We have known that these companies are losing money on every query made and that at some point they would have to actually start charging real money for these services.
Looking at the technology, it is still developing rapidly and the billions invested in hardware this year will need to be re-invested in the next generation in a few months, because the next generation is faster and more efficient... Didn't Microsoft have a warehouse full of nVidia cards they didn't have power to drive and by the time they got the power, the chips were so outdated (6 months), that it was cheaper to write-off the cards and buy new ones?
Somebody has to pay for this madness, and it shouldn't be the non-AI using customers of these services subsidising the heavy AI users. The AI pioneers have known they were getting a lot of usage for pretty much nothing, now the real bills are starting to arrive and they are "outraged" that they have to actually pay their way. If the AI is making their job easier and more efficient, they should be saving enough time and effort that the AI still pays for itself. If it doesn't, it is the wrong tool for the job and they should look for a more suitable tool that is more cost effective.
No different to the internet connected printers that phone home with page counts and requests for new toner from the leasing company. I suspect the same model, when the machine is getting empty, it calls for help and somebody comes out with the right selection of powders to refill it...
Only, it should never be on the office network, it should be on a separate network, just for such devices, or if that isn't possible, at least on the guest network, where it can't do any real harm.
At home, my "smart" TV from Sony (2015 model) stopped getting security updates in 2016. It hasn't been online since! I just plug in a FireTV or Apple TV, the Sony Bravia is still showing great images, it has been through 3 FireTVs and now an Apple TV 4K in the meantime. When the FireTVs stop getting security updates, they are just 20€ in a Prime Day deal to replace, the Apple TV a bit more, but still a bargain compared to replacing a 55" TV every 18 months, when the "smarts" stop being so smart.
Pricing and availability will also be very important. Even if the hardware isn't up to current generation nVidia standards of performance, having something that can inference at the speeds you need today for a fraction of the cost of a new nVidia set-up, which will be delivered some time in the next 6-18 months, is going to be attractive for many.
When I look at current lead-times for non-AI servers, let alone those with Blackwell AI units, with delivery times of around 6 months not uncommon with some manufacturers, having a complete data centre up and running in a couple of months is enticing...
Yeah, we bought 6 servers with 2 Xeon Gold processors, 1TB RAM and 2 SANs with 60TB SSD storage for our main data center in October. I am now looking at updating another site with 4 servers, 1 Xeon Gold and 512GB RAM and only 1 SAN will cost more than the hardware for the main data center! Luckily the hardware on that site is not overly stressed at the moment and was coming up with the normal replacement cycle, so we are looking at extending the support another 2 years and see what happens to the prices in the meantime...
Most companies have policies in place and the training is to see if the users have read and understood the policy - such as not uploading company documents into AI systems...
Anyone caught using AI before they have been trained, for business purposes / on a business device, would face disciplinary action. After the training, if they are caught using a non-approved system, they face disciplinary action.
Breaking things is the one thing you can't do, when it comes to power... I worked on projects for nuclear reactors back in the 80s and early 90s, the testing involved is way ahead of what we were doing for other projects - and the level of testing done on normal projects has been mainly removed from the product lifecycle these days. We used to say that analysis and design was 30%, testing 60% and the rest code development, for power projects, the certification comes on top of that.
It isn't just the engineers, it is also the customers, they see a "working" prototype and assume the system is nearly finished and they want you to carry on working on the prototype, no matter how hard you argue it is just a mock-up prototype and needs completely re-writing, they insist that the prototype is developed further. I've been called in on a couple of projects that went that way, it is never a pleasant experience.
Tell me about it. I was brought onto a project to help out. It was supposed to be a forecasting project in C++. The consultant did some mock-ups in Lotus 1-2-3 and the customer loved it, then said, "all our sales people have 1-2-3 on their PCs, just continue using that and export the results into the Oracle database. No matter how hard the consultant and project manager refused and said that it was just a mock-up and it would have to be re-written in C++, the customer dug their heels in and that was that...
Fast forward 2 months and I was brought on board to help with the 1-2-3 macros to automate 600 worksheets! It did sort of work, eventually. Then we started testing it and it was doing random things. We would step through the macros and it worked perfectly, run the macro and it didn't. In the end, we went to Lotus support, they asked asked for a copy of the "worksheet". They looked at the 5MB of worksheets (this was in DOS!) and their reply was along the lines of, "J**** H C****, 1-2-3 was never designed to do something like that, no wonder it doesn't work properly, re-write it in C++ for God's sake!"
Because the worksheet was so big and a lot of the macros used references, so would change on the fly, they were becoming corrupted in memory, because it was pushing the EMM365 memory boundaries in ways 1-2-3 had never envisaged. The code itself was an absolutely stunning demonstration of exactly what 1-2-3 was capable of, but it was also an absolutely stunning demonstration of what happens when you push its limits too far!
When I was a kid, my mother's boss had a degree in geology, but he couldn't write a cheque or drive a car... My mother used to write (and sign) all of his business and personal cheques for him.
He did eventually pass his driving test, bought a Porsche and then did a trackday at Silverstone and promptly drove it into the nearest crash barrier!
I've used the high pressure hose from a welding shop, but it was the pure air one, and I got the correct adapter off the production manager to use it as an air gun. Luckily the workshop had an outdoor area and I used that, the amount of dust it shifted was impressive... BUT, I made sure it was just air and I the nozzle a fair distance away from the motherboard, to test it first. I also use pens or screwdrivers to block fans, so they couldn't spin to excessive speeds and break the bearings or mountings.
It worked really well. But I would never grab a random hose and start spraying!
It comes out of my annual budget, and it reduces the profitability of my employer, which means the chances of a payrise are reduced and the chances the company goes bust, because their prices become uncompetitive increases. I am pushing PC upgrades for 2026 as far back as I can, in the hope that prices will stabalise and come back down a bit.
I bought a Sony Bravia TV over a decade ago. It got Android security updates for about 6 months, since then it has been offline. I certainly wouldn't buy another smart TV, at least I would never put it on the network. Since Sony stopped providing updates, it has been working fine and I have used 3 FireTVs and an Apple TV on it. The FireTVs probably aren't that much better and Apple won't be perfect either, but at least they will be (relatively) secure and up to date.
I really don't see the point in providing the "smarts" in the TV itself, you have a component that will probably get 18 months of updates, from launch, for a device that should last at least a decade.
I also used a PiHole, I recently switched to NextDNS so that I have DNS blocking on the move as well, blocking most tracking sites, but most people aren't in a position to be able to do that for themselves.
We need to stop this with regulation - which is why the appalling verdict in Italy just before Christmas was so bad (let third parties track iPhone users, because Apple has the ability to as well, instead of forcing Apple to let users restrict tracking on their own apps).
When I started working in IT, when we came up with new systems, we also had to have a fully working manual process that could be implemented in the event of the systems not being available... Today, not so much.
Times have changed and, instead of computer systems augmenting processes, they are the processes. On the one side, it is a win, you have all your books on a simple tablet or ereader and the kids don't have to lug around kilos of books and swap them out between classes in their lockers, on the other hand, if the systems are hacked, you have no backup.
It is also the other systems that are a problem - attendance records etc. taking a register with pen and paper was always something I loved watching as a kid. I loved the symmetry of the slashes for each pupil in attendance. With it all being electronic, you have to have some sort of fall back and then a way to get that manually collected information back into the online system, when it is up and working again.
That said, I don't see it being that much of a problem to provide ad-hoc classes with pen and paper, I've had a few occasions, where I had to give a presentation and the hardware failed and I had to give the lecture or presentation from my notes and memory.
When one of their servers DOSed our Internet connection, I tried abuse@ and postmaster@ addresses, both came back with "we receive so many emails that we delete them without reading, please refer to the section of our website that deals with your problem". I tried calling, I got the same message, but verbally and then got cut off... Funny, I couldn't find the part of their website which dealt with one of their servers going rogue and DOSing our line...
Even contacting their Twitter account didn't help. In the end, we had to get a new IP-address. Luckily we were in the process of switching providers and simply accelerated the switch. When I checked back 3 months later, when the contract ran out on the old line, it was still being bombarded by the Google server - according to the ISP, they were trying to stuff 1gbps down the 10mbps line! (The old ISP did offer to set-up their DOS protection, but it would cost us several hundred Euros a month, which we didn't want to pay, as this was Google's problem, not ours...
Having worked as a translator and having tried to use AI to do translations, I can say that AI still can't do translations solidly enough. I wasn't fully trained, I worked as an intern at a local translation company and, although my translations were readable, and way better than AI, I was nowhere near good enough to become a professional.
Even now, a few years later, I have to laugh at a lot of translations that AI throws up. It seems to have real problems going between German and English, for example.
One big problem, which now seems to have been solved, was using formal English would "fool" the AI. saying "do not" would have the AI dropping the word "not" from the translation ("do not open the case, no user serviceable parts inside" became "Gehäuse öffnen, nichts drin" (open the case, nothing inside), not exactly what you want the customer to read, if they have forked out 5K an industry terminal). Use "don't" and it would translate correctly.
Now imagine being at the zoo and seeing a sign that said "do not enter the enclosure" and the AI translation app dropping the "not" from the translation?
If I am writing a manual and it drops the "not", I can have a good laugh about it, if tourists start wandering into enclosures with lions and tigers in them, because the translation app said they should, that is another matter entirely.
This is why you need a knowledgeable human in the loop, to sign off on these things.
It is from before AI, but with WIndows Vista, Microsoft decided that the workers in Redmond could do a better job of the localisation than the German office, so the properties for network neighbourhood became something like "change your neighbour's nature/character". Unfortunately, it never did do what it said it would do, unfortunately. I had to wait until they moved out...
AI will no doubt improve over time, but if you don't have a knowledgeable expert to double check the results, how will you ever know if it is right or wrong, or if it is improving or getting worse?
When I qualified for a company car, back in '98, I looked for a car and applied to the fleet management, they asked how many miles I drive (standard for the company cars was 60,000 over 3 years). I told them I do around 60,000 business miles, with another 15-20,000 private (and around another 20,000 on my motorbike - I did a Euro tour each summer, which was around 12-14,000 over a week and a half or so on the bike). They said that I couldn't do that, that was too many miles, I pointed out that I had been doing that for the last 15 years, a majority of it in my private vehicle, so, yes, I can do that!
I have no idea how much it cost them, but they did finally get the car for me.
I was doing something similar in the UK, around 70,000 miles a year business, with my private car. The company offered a leasing offer for employees, starting at 99UKP a month, I phoned up for a quote, they couldn't give me one on the phone. They got back to me, for a 15,000UKP car (mid-range at the time, end of the 90s), they wanted 3,500UKP a month for the lease! Luckily, at the time, they didn't want to know mileage when taking out insurance.
Not monetarily expensive...
I worked for a company that printed money (literally). Governments would get them to design and print their currency, if they didn't have their own mint. They also had a side-line in ID cards and elections.
They were hired to run the elections in for an African country that was having problems with rebels at the time, luckily I didn't have a passport, so I was stuck in head office providing support.
We used Lotus cc:Mail and dial-in modems. Everything was working fine, until the team set-up shop and tried to contact the mail server. They kept complaining that it wasn't working. In the end, I plugged a phone into one of the modems and got them to call in and listened on the line. In the middle of the modem handshake there was a loud click in the line and both modems dropped the connection. The government was attempting to tap the line, but their equipment was so old, it made an audible click on the line as it kicked in. The team had to formally request that the line not be tapped and that it was only used for computer connections, which they couldn't listen in on anyway.
The same operator once caught me away from my terminal. Policy was to log off when you left your desk, I was just running around the corner and in the middle of a big edit, so I left the terminal logged on.
When I came back, he had made some white space in the middle of the file and prominent in the middle of the screen was the sentence "write out 1,000 times, I will log off my terminal, when I leave my desk!"
So, I opened a new file, wrote the sentence and copy and pasted it 1,000 times into the file (macro). I then exited the file and started the VAX Phone utility (a forerunner for ICQ and everything that came after it). The operator answered and I piped the file to his terminal! :-D
I once met a senior manager who was on the receiving end of something similar. A secretary at one of his customers was working on an email to send to the SM, she left her desk to carry out some task quickly, came back and sent the email... Only, while she was away, some joker had written some inappropriate text in the middle of the email, along the lines of a nice rear and wanting some hands on experience with it... The poor woman was mortified, when she found out, but, luckily the senior manager at the supplier thought it was absolutely hilarious and was laughing and telling everyone about the email he had received, so there was no damage done to the relationship between the two companies. I don't know hwat happened to the idiot that typed the message...
But another time, I was working late, along with a couple of other people, when the site manager came in and ordered everybody to leave. I told him I had a deadline, he said, no problem, he'd talk to the customer, leave NOW!
I found out later that a colleague had gone up to the his PA, opened his trousers and plonked the contents on her desk and asked her, what she could do with that... She said she needed a second opnion and called her manager... That was why we were asked to leave, he was marched in, once we had left, to clear out his desk, never to be seen again. I have no idea how he explained his sudden lack of a job and, probably, impending prosecution, to his fiancé once he got home. Again, not me, but that was a very expensive mistake for the colleague, and I would assume devastating the fiancé and traumatic for the PA.