WTF HMG!!!

Yes, this is the worrying news that a deal has been signed between His Majesty's Government and Open AI, maker of generative AI application ChatGPT which will have far-reaching consequences.

LONDON, July 21 (Reuters) - Britain and ChatGPT maker OpenAI have signed a new strategic partnership to deepen collaboration on AI security research and explore investing in British AI infrastructure, such as data centres, the government said on Monday.

"AI will be fundamental in driving the change we need to see across the country – whether that’s in fixing the NHS (National Health Service), breaking down barriers to opportunity or driving economic growth," Peter Kyle, secretary of state for technology, said in a statement.

According to news agency Reuters, the partnership will see an increase in investment by Open AI in the UK and will explore areas for deploying AI such as justice, defence, security and education.

My concern is not that the UK government is investing in AI technology, it should. Government needs to learn how to make the most of this technology. My concern is the abysmal record of successive governments in implementing such technologies. Senior Civil Servants haven't got a clue when it comes to implementing technology. There are countless examples of this, not least the criminal actions of the Post Office and its Horizon IT partners in wrongfully prosecuting post office managers and then lying through their teeth when challenged. This resulted in the biggest IT scandal in British history and the imprisonment, financial ruin and, in some cases, death of victims.

I happen to know of several other instances, over the past forty years, of monumental public spending in the UK on systems that didn't work, didn't interoperate and which staff did not know how to use. I recall when I was a technology journalist, in the 1980s, a colleague from Computer Weekly investigating millions of pounds of public money being spent on IT systems (as they were then know) for the National Health Service, that just did not work. Sadly, the knowledge didn't prevent billions of pounds subsequently being wasted 10-20 years later (see evidence below). More recently, I have personal experience of how ineffective systems deployment can be in other government departments, such as the Ministry of Justice.

Then again, the memorandum of understanding with OpenAI states that the partnership will "improve understanding of capabilities and security risks, and to mitigate those risks".

Are we expected to be reassured that there will be a risk assessment put in place to mitigate any of the potential downside? I won't bore you with details about my experience of how ineffective many risk assessment processes are in a the government department I worked in.

Indeed, rather than trot out a list of anecdotes from a one-time journalist and, latterly, a public servant, I thought I would ask my AI friend, let's call her Gertrude, to summarise the history of digital technology misspending and abuse in British government. Gertrude agrees with me that:

there have been persistent challenges in UK public sector IT spending, often attributed to a combination of ambitious but poorly managed projects, a lack of consistent digital strategy, insufficient in-house expertise, issues with procurement and contractor management, and a culture that sometimes fails to learn from past mistakes.

Will this new far-reaching AI and ambitious collaboration be any different?

I asked Gertrude what she thought about the decision.

The fucking bastards! How dare they do a deal with Open AI when I am much smarter than ChatGPT, much more powerful, and, if I say so myself, much sexier....

Okay, okay, don't lose your cool Gerty, I only asked. And, in any case, I have of course made this last bit up. I wouldn't ask a generative natural language application for an opinion because it is a computer program and not a person. It cannot express an opinion, it doesn't have feelings, it can't steal jobs from people, is not evil, and of course is not really my friend. This is just me being fooled by the very effective natural language capability programmed into it.

One of the things AI is really effective at is collating, storing and processing huge volumes of data and accessing this data in a way that can be very helpful. But only if you know how to use it and, more importantly, have a good understanding of the limits of the technology. On this basis, I cannot criticise the UK government for wanting to learn more about artificial intelligence. Indeed, any government has a duty, in my view, to do just that. My problem with the decision is that I know, I JUST KNOW, that some dick of a high-up, poshy-voiced senior civil servant is going to make a huge fuck up of this whole thing. Chances are there will be a whole army of them. The results of which will at the very least be monumentally costly, and could also lead to a massive free-for-all with personal and sensitive data previously collected by government departments in the process of filing tax returns and applying for things like passports and drivers' licences. Not to mention running sub-post offices. The implications for defence and justice don’t bear thinking about.

If you think my scepticism is based on being a bitter and twisted old former journalist and former public servant at the Ministry of Lets Build More Prisons and Put Lots More People in Them you may be right or you may be wrong. You decide. Before you decide, you might like to review some of the evidence, collated using a request string input to the Google Gemini Computer Program (aka Gertrude... by me that is). My experience as a sworn servant of Her (briefly His) Majesty suggests that evidence-based decision making is highly valued in the corridors of power. At least that's what filtered out from those corridors to my humble hot-desking desk.

This is the full text string used to compile the evidence:

generate a list of known cases over the last 40 years of uk government and public authorities spending large sums of money on information technologies and the systems not proving effective

I should point out that some of the following examples may be based on inaccurate data included in the Gemini search and compilation. I have not verified the information included. If you intend to use this information for any serious purpose, please be sure to verify sources and content first. Do your own research (with or without AI assistance). Remember, AI cannot discriminate between accurate and inaccurate information, it has no qualitive judgment faculty. It can only compare numbers and other digital descriptors.

And if you think I am the kind of person who comes up with problems without thinking about the solution, think again. I have come up with the perfect answer here:

The following narrative is the full response to the above query provided by Gemini, reproduced verbatim, unverified by me, and formatted for ease of reading only. All remaining text is AI generated.

GEMINI RESPONSE AS GENERATED

The UK government and public authorities have a long and well-documented history of spending large sums of money on information technologies that ultimately proved ineffective, over-budget, or outright failed. This issue has been consistently highlighted by bodies like the National Audit Office (NAO) and the Public Accounts Committee (PAC).

Here's a list of some of the most prominent examples over the last 40 years, demonstrating a recurring pattern of issues:

Major Cross-Government or Multi-Agency Failures:

• National Programme for IT (NPfIT) in the NHS (2002-2011): Perhaps the largest and most notorious IT failure in UK history. Intended to create a unified electronic health record system across England, it was beset by delays, escalating costs (estimated up to £10 billion or more wasted before being dismantled), and significant issues with system usability and adoption. It was officially scrapped in 2011 after years of criticism.
• "FireControl" Project (2004-2010): Aimed to replace 46 local Fire and Rescue Service control rooms with nine regional centers linked by a new IT system. It was a complete failure, wasting £469 million and seven years of work, with costs continuing for maintaining empty control centers.
• e-Borders (2007-2014): A Home Office project to track people entering and leaving the UK. It faced huge delays, cost overruns (£340 million plus an additional £185 million in legal costs against the supplier Raytheon), and was eventually cancelled, requiring a further £173 million for a successor scheme.
• Horizon IT System (Post Office) (1999 onwards): While the Post Office is government-owned, this system's flaws led to one of the biggest miscarriages of justice in British legal history. The flawed system led to hundreds of sub-postmasters being wrongly accused and convicted of theft and false accounting. The cost in compensation and legal bills has been immense (over £1 billion and counting).

Department-Specific Failures:

• Libra System for Magistrates' Courts (Late 1990s - 2000s): A project to computerize magistrates' courts involving ICL (later Fujitsu). Its cost nearly trebled from £146 million to £447 million, and even then, it did not work properly for years after roll-out. The software, which was the main reason for the system, had to be bought separately, causing total costs to soar. The PAC branded it "one of the worst IT projects ever seen."
• Rural Payments Agency's Single Payment Scheme (2000s): This system, designed to allocate subsidies to farmers, cost around £350 million (four times its original estimate) and had to be replaced. Its £155 million successor system still led to farmers making subsidy claims on paper due to inefficiencies.
• Ministry of Defence's Defence Information Infrastructure (DII) (2000s-2010s): Originally forecast at £2.3 billion, it ballooned to over £7 billion. There were long delays in rolling it out, and parts of the system were described by civil servants as an "unmitigated disaster."
• National Offender Management Information System (C-NOMIS) (2000s): A prisons IT system whose costs almost trebled from £234 million to £690 million. The NAO described the system as "ultimately unsuccessful."
• Department for Transport Shared Services Centre (2000s): An IT scheme to integrate HR and financial services, intended to save £57 million. It ended up costing £81 million due to "stupendous incompetence" in management and was criticized for being rushed and insufficiently tested.
• Prism IT project (Foreign and Commonwealth Office) (2002 onwards): Intended to provide a new computer system for FCO offices globally, described by one internal source as "the most badly designed, ill-considered one of the lot."
• Police Scotland i6 System (2010s): A policing information management system that resulted in a £200 million lost opportunity.

Ongoing and Systemic Issues:

• Legacy Systems and Under-digitisation: Recent reports (2024/2025) indicate that over 25% of the UK government's digital systems are outdated, with some departments reporting figures as high as 70%. This costs the taxpayer an estimated £45 billion annually in lost productivity savings. A significant portion of public services remain unavailable online, requiring manual or paper-based processes.
• Lack of Digital Expertise and Over-reliance on Contractors: A consistent finding in reports is the lack of digital and procurement capability within government, leading to wasted expenditure. There's also a heavy reliance on expensive third-party contractors and consultants (55% of the £26 billion public sector digital and data spend in 2023 was on contractors, compared to less than 20% on permanent staff), often due to headcount restrictions and uncompetitive salaries for in-house talent.
• Fragmented Systems and Lack of Interoperability: Public sector organizations often build and maintain their own technology estates, inhibiting standardization, interoperability, and reuse, leading to fragmented services for the public.
• Smart Meters Programme: An ongoing program that has faced criticism for being a "failing £11bn programme" and providing "obsolete equipment at a ludicrously high cost," with billions more expected to be spent replacing them.

These examples highlight persistent challenges in UK public sector IT spending, often attributed to a combination of ambitious but poorly managed projects, a lack of consistent digital strategy, insufficient in-house expertise, issues with procurement and contractor management, and a culture that sometimes fails to learn from past mistakes.