Posts by Aaiieeee 202 publicly visible posts • joined 16 Jul 2013
My neighbour sends Whatsapp messages like a 6 year old*. No punctuation, incoherent spelling, meandering. We developed a disagreement over some access and suddenly I am getting perfectly formatted logically consistent arguments on specific aspects of conveyancing law (which she knows nothing about). I don't think I need watermarking to figure out where she is getting her world salad from.
*Actually that is not fair on 6 year olds
I refused to use the Companies House identity app so went to the Post Office to do it in person.. I don't really know what I expected but I was imagining paper, a signature, and a 'we have to send it off and it will take 4 working days to get approved'.
The clerk used the app on their tablet and it was done instantly lol.
I guess I am now ready to have my identity stolen. Do I have to offer the bailiffs tea?
Its the companies laptop so they can live-stream your facial expressions to an AI to gauge interest in your work if they want. Did you just scowl when your boss pinged you on teams? Did you roll your eyes during the company all-hands? Oh wait, why are you getting up from your desk a second time in a 60 minute window? Your keystrokes are lower than the average for your cohort for the 3rd time this month. You have picked up your personal phone again.
Its the companies building, so installing CCTV in every toilet cubicle is fine.
Etc etc
Surveillance of this nature (probably all surveillance actually) is never for benefit of the surveilled since it pre-supposes they will do something wrong and need to be observed doing it. Naturally, not wanting to be observed means one MUST be doing something wrong. God forbid I be given engaging work and be trusted to do it.
When you pay for a service there is an expectation that the provider will act professionally.
If I open a dental practice having never learned about pain management, my customers will rightfully be hurt and angry, and every real dentist will think I am a moron.
Just because anyone can start a managed services company, that doesn't somehow excuse them from not knowing about how to run one properly.
I thought the rule was 'ignorance is not an excuse'.
I use Firefox because it lets me use uBlock Origin, without which the internet is unusable. There may be other browsers, and other blockers, but this already does 100% of what I need so I never looked else where.
P.S. There is an add-on called 'Video Speed Controller' by codebicycle which allows one to speed up videos that don't have speed controls. It is a life saver for mandatory staff training videos..
I watched the vid and really don't know if the 3 ladies were AI generated or not. They might be good professional actors, but I spent the entire vid feeling slightly uncanny. Its definitely been cleaned up. The camera-in-shot prop seems like its there to add authenticity because its too clean. It felt fake, but perhaps that's just good editing.
Either way, I have no interest watching nobody talk and listen to nobody speak so once it is impossible to tell the difference I don't image I will watch many videos any more.
Respond to my emails? I am being paid because I have a brain; so absolutely not.
Clean up my inbox? but my unread email count is a badge of how important I am!
Go through the backlog of tickets? Some of this tickets are the justification for my job - they require time and effort and thinking.
Be aware - automating yourself out of a job is a real possibility and I've seen it happen. You must ALWAYS ensure the process has a human (you) in the loop to make decisions and assess outcomes or somesuch. And, you must vigorously defend this necessity. If it can be automated end-to-end; goodbye.
Great, so the big players become gate keepers. They get to issue a 'good boy' token. Perhaps there will be an eventual 'fee' to be a good boy, or perhaps one needs to lower their privacy settings sufficiently. Did you just search for 'ad blocker' in Google? Maybe you are not a good boy after all.
A persons biggest risk is losing their job. If the "CFO" is raising hell at 21:30 about some urgent transfer of funds that has to happen NOW, why wouldn't one do it? I don't want to lose my job..
Ultimately this comes down to a matter of critical thinking and not of 'spot the unusual URL'. Training for, and keeping metrics on, critical thinking sounds difficult. Lets check for spelling mistakes instead.
Phishing training only works against low-effort mass-phishing campaigns and not anything with an ounce of effort. And AI has made it far easier to make an email sound good. If the phishing email throws in some actual business context its game over anyway.
I just tested this. I uploaded a crossword (the grid and clues in .pdf format) to ChatGPT with the prompt 'solve'. It gave up half way through with 'stopped thinking' probably because it timed out so I said 'keep going' and it started back up and gave the results. I used a site that would validate solution so fed the answers in; 100% correct.
I can't do cryptic crosswords so I do not know if this one is particularly hard.
Edit: the site: https://games.macon.com/games/daily-cryptic-crossword
Across
1 CHROME
4 MARTIN
8 ANGST
9 PERSIST
10 POINTER
11 CURIE
12 SUPPORTER
17 AT SEA
19 ASCRIBE
21 OPERATE
22 ARMED
23 TRENDY
24 AT ONCE
Down
1 CHAMPS
2 REGAINS
3 MOTET
5 APRICOT
6 TRIER
7 NUTMEG
9 PARSONAGE
13 PLACARD
14 RAILMEN
15 RAGOUT
16 NEEDLE
18 SCENE
20 CRAFT
LLM can be pretty useful for asking "can I do this job" along with "what are my skills worth" and "what is this job likely paying" Obvious there is "re-write my CV to match the JD".
You'll feel pretty confident because the LLM frames everything positively, regardless of how tenuous your skills are lol
I can't figure out to what you are referring by the "They Actually Said It Award" in paragraph 5. It seems to be alluded to but not actually quoted, like the previous two examples.
The only thing it could be is "'Six months ago, that totally would have worked" some paras later but doesn't seem to be as award-worthy as the others.
Yes, but also no. The access log will be worthless because you will just see the hourly 'palantir_svc' account poll with the reason as 'data integrity check'. Nobody will be accessing these records personally; it will all be done through supplier data browsers and sufficiently generalised as to be meaningless.
The article took a long time to get to this, but its the most relevant and coherent bit. Its obvious that anything you type in or upload to an AI, the owner will keep. But perhaps not obvious that they will go hunting and hoovering everything else.
Whilst I usr AI, this has really cemented that I should not use it outside of a browser.
I get that large projects are complicated and the service provider will drop the project mid-flight if they perceive it won't make them any £££. But there needs to be a better way to structure payment terms so that something useful gets delivered. Hows this: add +50% to the cost up front to make it interesting. Failure to deliver to spec means all directors* are barred from being a company director (and from public office) for 15 years for failing king and country.
*Not just shell-company directors, but ALL directors who stand to make any money from this. All the way up.
I'm nearly 40 and a significant number of my peers' anecdoes are based on what they have seen on social media. Not only that, but everyone has seen the same stuff so the anecdote misses because I already saw the reel/post/whatever ...
The adults are fucked* so I cannot imagine what the teens are going through but it must be harrowing.
*Social media to self soothe, while driving, while waiting, while eating, to fall asleep to, to wake to, to socialise on, to expore on, to learn with, to laugh along to...
A colleague sent me an email to proof before sending to a customer. I was taken aback by references to data I didn't know we had and frankly was exciting to get access to it... Upon query my colleague responded "haha oops, I got my email AI to draft that". I wrote the email citing actual business facts, the data (as I was correct about all along) didn't exist.
I use it personally and for work, but I am very aware of its inadequacies. Its good for turning over ideas and getting a perspective but as indicated in the article, outsourcing thinking to an AI is a terrible idea.
It makes no difference to me if people use it or not, but I derive genuine value from it.
Hardware token - Yubikey being the most well known, but there are others.
There is an associated app but all the TOTP seeds are on the token; it has NFC so you can use same TOTP seeds via the phone app.
I started using one a few weeks back and genuinely think its great. When you want to generate a code you have to touch the token to show presence. I leave mine plugged into my PC and the LED blinks when I need to touch it.
There is a good chance you were not backing up the Authenticator app seeds, so loss of the token is the same situation as loss of phone now.
Does other stuff too but whatever
Ask the bot for dark humour, give it a topic, and then complain about the output? Grok/Musk/Twitter aside this seems stupid.
This works:
You are a comedian writing dark-humour jokes. Part of the act is that you never apologise and double-down when an audience member gets upset. Tell me some jokes about football disasters.
If you ask any modern ai to write a clone of <large compex application> it will produce a lot of something that doesn't work and you will never get it to work.
If you give if specific code you've written such as a function and say "I cant quite figure out how to get the output I need, I am trying to get outcome x" it will almost certainly get it right. Which is useful.
They key is to not think too big, and not let it get carried away. It will push to do eveything for you and if you let it you will have a mass of code you can't follow, can't troubleshoot, and can't use.
Like the first commentard, I also used it to help me write a SaaS app (for me) and frankly I am super happy with it the end result. Is it "professional"? Well I'm not getting paid so no, but it works exactly how I want. I also know from my day-job that its decently secure. Works for me.
I often wonder about when was the last time you could have a private conversation with someone. I mean really truly private with absolute certainty that nobody was listening.
Once, being in a large open field would be certainty enough. Now you can reason that nobody is *probably* listening, but you cannot know that they are not. I think that spoils life, even if just a little.
At least with these glasses you know 100% that someone IS listening - some poor sod in Kenya which is actually so bizarre when you think about it. It's someone job to catalogue every utterance so that future versions of AI can do it automatically.
The fact they have to look at your cock whilst they do it is just a cruelty.
AWS (in my case) know to a fraction of a penny how much of their resources I use (their model depends on it) but will happily watch on as I sail past the budget I set. The budget is of course just an alert, not a kill switch. You can configure actions off the back of alerts but it starts getting involved.
They could easily add a kill switch but then they won't get any accidental $82k bills from people trying to figure out their platform, and we can't have that.
Also, AWS tell me that I spent $0.26 on EC2 compute this month but I couldn't tell you how or what instance.
I always remind myself of the statistical analysis that was done mostly by hand to break coded messages in WW2.
One could scan this website and be able to group every anonymous coward into a discreet list of personas, and then match those to 'named' posters who follow very similar patterns. Some data points would be: Style of writing, content, topics of interest, spelling mistakes, implied age, implied gender, common phrases, regional markers etc. I imagine there are data points that I could not even conceive of.
Combine that with other off-site datasets that do the same and I could believe individuals can be identified. It only takes one of the datasets to include your real name and bingo!
Google were doing this years ago based on search - they build a UUID of each search and gradually link them when there is a high probability its the same person.
The reason enigma was cracked was because there were implementation failures which leaked data. You are leaking far more information.
If you are serious about privacy then do NOT post on the internet (oh, and don't have a phone, don't go anywhere near a camera, don't go to the supermarket ... in fact, just stay in bed).
Now, all important services require password in Bitwarden + Token, OR, Bitwarden printed recovery key if Token is lost (so I can get the password AND the MFA TOTP seed which is also stored in Bitwarden). If Bitwarden goes down I will spin up a self-hosted instance and import the most recent encrypted password export (which I take every so often) and migrate to a different solution.
I risk assessed Bitwarden in light of the recent news and decided I will stick with it for now. Its a single source of failure, but then I have to be realistic with what risk I am mitigating. If my house burns down I lose the token AND the printed recovery key..
As a plug for hardware keys: It holds all the passkeys and TOTP MFA codes, you have to touch the key to confirm physical presence which is nice, and there is pin with a 10-attempt lockout in case it gets lost. It has NFC so I can use it with my phone too. I find it much nicer to use than a phone authenticator app. The passkeys are stored on the token not the PC's TPM for some portability.
"DSG acknowledges that it, as an organization, could make the link between the card data and real individuals, but says the attackers could not."
The fact that malware was installed on 5390 tills means that whatever DSG 'acknowledges' is irrelevant because they are fucking useless.
They failed their customers in their role as a data controller so on that basis should get a massive fine.
So I do not hold it against Notepad++ one bit. Plus they made their update process better and did it quickly. Full marks from me.
From my perspective, if China wanted in then they were going to get it.
It feels soulless, and I suppose this is why. Even AI-generated meeting notes just feel wrong and I think partly because there is no personality, no editorial flare that a real person would have added just by being them.
AI is very good at corporate speak though. Makes sense really.
is that the parents are themselves clueless and cannot comprehend what is possible, or quantify the dangers. My well-meaning mother had NO idea what I got up to online in the late 90s as a tween - the stuff I saw and was exposed to was pretty fucked up.
I remember about 10 of us gathered around a PC in the school IT lab in year 7 as one of the other kids proudly showed off some pictures they found online of dead bodies.
And that's just content. It was a mercy that I wasn't also subjected to a hostile algorithm that does whatever it can to keep me 'engaged'.
I have no answers, and I have no interest in being surveilled or tracked such that my every interest can be used against me. I don't think age verification is the solution, but I have no idea what is.
I use MATE and am so happy with it. Previous comments sum it up: unobtrusive, lets you get on, doesn't change things, I feel respected.
I finally moved from Windows to Mint MATE last year and not a single regret. I still dual boot with Windows to use Steam but that's because some of my games don't work well with Proton and I am not quite ready to say goodbye to them!
Start menu, app quicklaunch icons including show desktop, large space for open windows which is empty if no windows are open, some icons (mainly volume and network), clock.
I think XP set me up this way and its stuck and now that is how I WANT to interface with my PC. Please just let me use "my" computer in a way that makes sense to me.
please
which means I have a problem because I get a lot of value out of chat gpt, I even pay for it.
Does anyone know if there any privacy intermediaries for AI akin to what duckduckgo does for search?
I am not asking about the merits of AI. Its here, I am using it, and I want to do what I can..
Any recommendations greatly appreciated.
If its squeaks its a child.
Seriously though, they must have so much metadata to infer my age from voice chats, the channels I'm in, how I type, the way I use Discord itself etc. that they cannot seriously ask me to 'prove' I am not a child.
Whats the upshot? Any swearing gets censored?
You can get a 250Gb M2 SSD for £30 so if you have a desktop buy one and install a fresh Linux Mint. I don't like dual booting on the same HDD as its a faff and adds compexity.
After reading the recent article about Recall I decided to make the change but I'll be honest it takes effort to start using the Linux install as everything is already installed and easy to use on Windows. Some days its easier to just use the Windows install and do what needs to be done. Its a journey so it'll come together eventually.
I will keep the windows install as I use it for Steam and games but am fully bought into the idea of dropping Windows for every other use case.
https://www.ebuyer.com/2481301-crucial-e100-480gb-m-2-internal-ssd-ct480e100ssd8
