So they want to create an endless trove of data on your PC that might, somehow, someway be useful to you?
Why? What's in it for them?
Frankly this seems wasted outside of a corporate environement, and even then it's fraught with risk
230 publicly visible posts • joined 12 Jun 2009
It was finding Co-Pilot and Search with Bing in Notepad of all places that made me go "wtf!"
I can just about see a case for it being built into things like Outlook and Word given that the most likely use case for LLM's is helping compose or re-word emails and documents (not that most want to do that!)
Lucky Registry hacks still let you get around it and revert to classic Notepad but still....
What shocked me most about this relentless push?
Finding Co-Pilot and Search with Bing inside NOTEPAD
Yes, the application designed specifically to let you quickly tweak configuration files, had frigging Co-Pilot and Search functionality, what's more, it'd automatically signed into my Microsoft account.
A step for far, so I found a registry hack to let you go back to the old Notepad design.
So let me get this straight
They decided to take an operating system
That is part of some of the most dangerous weapon systems on Earth
Put it in a phone, which is online 24/7
and then throw down a challenge to the entire fecking internet to prove that their OS is "unhackable"
No.... I can't possibly see where this might have knock on effects.....
Either we get more frequent desktop refreshes driven by the need for an increase in power required to perform everyday actions; or we have the majority of day to day business activities in the cloud. The latter *should* require far less local resources required on each PC because the cloud servers are doing the heavy lifting.
Despite the continual money grab, it's the one thing you can say that's positive about the software as a service model, the continual cashflow gets over the resistance to continual patching, and in turn allows you to stay current on security patches.
Trying to get SME businesses (and we're talking the very S end here), to update software just because "It's now out of security support" and "You risk bigger fines if you have a data breach" is nigh on impossible and almost none of the SQL new features really matter to the existing use cases. There's no "spend some money, it'll run faster" or "spend money and it'll do this thing you wanted that it couldn't do before"
LLM's on a PC will be best placed to helping people who don't actually understand technology, and in turn mean that for those of us who do understand it, UX/UI designers will likely put even less effort into resolving it.
Think about the slow creep from Windows XP/7 era Control Panel to Windows 10/11 settings app, many functions get migrated across, but not all of them, leading to a confusing mishmash of where to go to fix a problem.
AI will add another layer, because some of the bits will only be found by asking the AI where they are. Helpful for a non-techy who just wants to make the icons on the screen bigger because they're struggling to see things, but crap for techies
It's ironic really, as features go, the use cases for it (narrow as they are), belong entirely within the Enterprise sector.
As a compliment to the various Legal retention policies, searchable snapshots of employee actions that can be compiled and searched for efficiently is ideal for big business to ensure that their legal responsibilities are being met. Even more so if the entire policing aspect can be handed off to a properly trained LLM.
So yes, it's spyware, yes, it's horrific to even consider deploying this to private individual's machines, but as a properly controlled tool in environments where individuals don't have an expectation of privacy due to using a work machine, maybe not entirely useless?
The biggest problem with securing Microsoft products is the sheer depth and complexity that they have. Allowing for support from a one man band up to a 100,000 seat enterprise within a single security design schema is bound to cause problems.
For security to work it's got to be simple to implement, easy to interact with on a day to day basis, and easy to understand. Only a couple of thousand people in the world *really* understand Microsoft Entra and Active Directory, and yet ten's of millions of people are there using it to "secure" their systems.
I think the answer is for the government to implement data object standards, ones that can describe all the most commonly described items that software has to deal with from employees, to claimants, to suppliers, to stock items, to asset management and so on. With data standards in place, with a level of flexibility, then software can be written to handle those objects through processes as required, and things can be shared between different software platforms without a lock-in.
It's no different to manufacturing using standard parts, if all your data components meet the specification, then you can use a variety of ways to put them together.
There's no question that the NHS needs more effective IT systems but it'll never get it done for a sensible cost whilst it keeps chopping and changing it's mind based on the whims of the political taskmasters. I'd argue that the NHS needs a committee of IT and medical experts to agree a data standard for patient records and communications, a standard as clear and as technical as those that under pin things like USB, TCP/IP etc. etc.
Once anyone and their dog can write software that's compatible and that makes it possible to interchange data, then the commercial markets will finally be able to actually compete with each other and deliver software that saves time and money. Yes, there'll be shitty pieces of software out there, but those will die out over time.
It's not hard to use ChatGPT professionally, you treat it's output like a search engine result, you ask for the citations and sources used to construct the argument and then cross-check the information yourself against both the cited sources, and similar.
AI generated content should be the *start* of something that you produce, not the entire product.
Unsurprised at the feedback, the idea that wearing your monitor on your head will improve things isn't enough for the majority of use cases.
What I'd like to see is some of the technologies used moved into markets where having information directly at hand is critical.
For example, A&E, being able to call (with prior consent) someone's medical records, their current observations, and intelligent warnings about someone's condition could offer a big help.
Another area would be car windscreens, merging in some of the self driving car's sensors into a HUD to improve road and hazard visibility, particularly at night when the glare of oncoming headlights can make it difficult.
AR and VR technologies definitely need to stop and think about the use cases before they release and develop the hardware, it's not enough to release hardware and hope someone invests time into it
Said it before and I'll say it again, the governments need to define data standards for records, health data, education data, welfare data, etc. all needs to have a common data standard that's flexible to cope with future changes, but robust enough to survive long periods without the standard being updated.
If they did that, then all the software companies wanting to tender for government business would have to work off the same approach when it comes to data interchange.
The law's not going to achieve a lot because you really have to target the toolmakers and regulate them to ensure that their products can generate "random" porn, but that it doesn't support "undressing" of uploaded photos and doesn't accept prompts that relate to real-world people.
Same for targeting the companies that host the material, make it difficult and expensive to host the material, require ID checks for uploaders etc.
No law will put the genie back in the bottle but you can make it as hard as possible for people to hurt others in this way.
IT service providers need to start carrying financial penalties as well for any software screwups, even if they're end-user driven ones. It'll reduce the likelyhood that anyone gets to "go live" without appropriate system testing, including potentially dual-entry and ensuring that the new software is actually fit for purpose, and that staff understand what the new system does and why.
Too many of these systems go into companies and local authorities without either side understanding each other.
It means that the adverts support all levels of the marketing funnel from the broad "raise awareness" starting point, through interest, desire, action, loyalty and finally advocacy.
The idea being that a fully integrated advertising platform would let you get a message out to millions, identify the hundreds of thousands who show an interest, identify the tens of thousands who have a desire, track them to your store to find the thousands who make an action to purchase, track the hundreds who are loyal to the brand, and finally identify the tens of people who are advocating for your product or brand.
The irony being that many news outlets will use a news wire source like Reuters for a lot of their content, often copy and pasting the articles automatically or quoted large swathes.
Thus scrapped content from the site *will* seem similar if it's been taken from another source using the same news wire.
Targeted advertising will only satisfy privacy concerns when you chose your own boxes to be put into via an active mechanism.
If you chose no boxes, then you get whatever random ads are thrown up.
Advertisers need to agree a gold standard of categorisation for adverts so that tools can be developed by any and all companies as required.
And to encourage people to categorise themselves (or accept infered categorisation), then offer cashback rewards in the same way as "loyalty" cards do.
They'd be better off working out how to take all the legacy equipment and optimise it's use. Could all that 3G gear be repurposed to ensure that low bandwidth traffic like SMS and Push notifications for OTP are routed via the legacy equipement which is more than capable of supporting them, whilst the more intensive content is routed via the 4G or 5G network as appropriate.
It wouldn't be impossible to have a device or chipset which load balanced the traffic appropriately over the correct network option.
Having worked with for an SME ERP software developer for the past decade and a half, I can safely say that this project will fail as I guarentee that there will be too many "key decision makers" who will block things that they don't like, and hold differing views to their equivilents in the other departments.
It's hard enough to get a company of 20 people to agree, but on this scale? It's impossible.
They'd be better off using discrete solutions and agree a data reporting standard, so that performance figures can be compared and shared easily.
Imagine a "Money Lock" account in the UK?
Not only is it nearly impossible to find a bank branch within 30 miles these days, you turn up and some little beauracratic tinpot dictator will decide that *these* particular bills aren't valid, as they're not paper copies, and that your ID is invalid as it doesn't really look like you.
A good rule of thumb for fair contract law is "would a multi-national corporation buy on those terms"
The answer is, no, no they wouldn't, thus mid-contract rises have to go away.
Which in turn means that we need to have two seperate contracts.
1) A repayment agreement for the handset cost
2) A service contract that lasts a fixed period of time with no price rises.
Start making everything clearer for the consumer, and also, start giving an actual APR for handset loans, so people can compare them against other forms of credit.
The government and civil servants responsible for making the policy document were entirely incapable of predicting the future, nor accurately assessing the time to complete the rollout of a project in an area where they have next to no experience. Never saw the point of smart meters, never wanted one, will always resist having one. After all, why on earth put something on the public internet for the sake of a tiny bit of data sharing?
Chances are high that this was high level corruption, boost shareholders values in various companies by pretending to give a damn about something.
It's not hard, instead of infering what people might want, ask them. Offer them reward points for being matched against advertisers terms. Offer them the option to add what your analytical engines guess, but to ignore some bits.
Millions will sign up willingly, and those that don't will just have to put up with irrelvant stuff.
Said it before, and I'll say it again, governments both national and local need a common data standard that'll let software vendors write compatible software. Coupled with a common processing langauge that'll let data be transformed and move around.
Whether it's health records, HR notes, social worker's case notes, police records etc. People designing software rarely have the relevant experience to write good structured data. Thus the structure needs to come from the people who do know.
Working for an ERP software developer, the main reason you do this isn't to recover the information, it's because an update is far faster than a delete action, thus for the most part you update the record to hide it when a user deletes things, and then later on run maintainance to clear out the fragments when it's quiet.
Not to mention can you imagine the mess the indicies would get into if you had users deleting records constnatly during the working day?
I wonder if they've worked out that a lot of people don't *want* smart meters?
Why do we need a meter that can be disabled via a computer, that's part of a hackable network using relatively insecure, outdated protocols (GSM / SMS)
Has no one worked out yet that Russia has spent *decades* building up their cyberwarfare capabilities for this kind of reason??
It'll be the same problems that afflicts many organisations implementing ERP, each department will have it's own requirements that either contradict other departments or pulls the implementation team in too many directions.
It's almost certain that most of the time has been wasted focusing on integrating legacy technology, so that the new technology can come online for one department, without the other departments shifting because they're not "ready" yet.
Saw a good use for ChatGPT via Twitter
Someone ran a landscape gardening business, but was dylsexic, and often losing out to others based purely on how their written work was percieved.
The person behind the Twitter account, helped them setup a routine whereby they could send an email, ChatGPT read it and re-wrote it out, and then sent a reply back to deliver a more professional sounding email.
That's the kind of good that this kind of AI can be used for, not for original thought, but by helping present those thoughts more coherently.
There must also come a point whereby it makes sense to fit a power generation or building heating system into the cooling loop, particularly if you don't just balance the waste heat over the datacentre, but combine it with the entire building (with emergancy area isolation!)
After all, we can use heat energy in a variety of ways, and convert it back to electricity, especially if there's a fluid medium involved.