Sign of the times: might ChatGPT re-invigorate GPG?

It’s important to keep finding errors in LLM systems like ChatGPT, to remind us that, however eloquent they may be, they actually have very little knowledge of the real world.

A few days ago, I asked ChatGPT to describe the range of blog posts available on Status-Q. As part of the response it told me that ‘the website “statusq.org” was founded in 2017 by journalist and author Ben Hammersley.’ Now, Ben is a splendid fellow, but he’s not me. And this blog has been going a lot longer than that!

I corrected the date and the author, and it apologised. (It seems to be doing that a lot recently.) I asked if it learned when people corrected it, and it said yes. I then asked it my original question again, and it got the author right this time.

Later that afternoon, it told me that StatusQ.org was the the personal website of Neil Lawrence.  

Unknown

Neil is also a friend, so I forwarded it to him, complaining of identity theft!

A couple of days later, my friend Nicholas asked a similar question and was informed that “based on publicly available information, I can tell you that Status-Q is the personal blog of Simon Wardley”.  Where is this publicly-available information, I’d like to know!

The moral of the story is not to believe anything you read on the Net, especially if you suspect some kind of AI system may be involved.  Don’t necessarily assume that they’re a tool to make us smarter!

When the web breaks, how will we fix it?

So I was thinking about the whole question of attribution, and ownership of content, when I came across this post, which was written by Fred Wilson way back in the distant AI past (ie. in December).  An excerpt:

I attended a dinner this past week with USV portfolio founders and one who works in education told us that ChatGPT has effectively ended the essay as a way for teachers to assess student progress. It will be easier for a student to prompt ChatGPT to write the essay than to write it themselves.

It is not just language models that are making huge advances. AIs can produce incredible audio and video as well. I am certain that an AI can produce a podcast or video of me saying something I did not say and would not say. I haven’t seen it yet, but it is inevitable.

So what do we do about this world we are living in where content can be created by machines and ascribed to us?

His solution: we need to sign things cryptographically.

Now this is something that geeks have been able to do for a long time.  You can take a chunk of text (or any data) and produce a signature using a secret key to which only you have access.  If I take the start of this post: the plain text version of everything starting from “It’s important” at the top down to “sign things cryptographically.” in the above paragraph, I can sign it using my GPG private key. This produces a signature which looks like this:

-----BEGIN PGP SIGNATURE-----
iQEzBAEBCgAdFiEENvIIPyk+1P2DhHuDCTKOi/lGS18FAmRJq1oACgkQCTKOi/lG
S1/E8wgAx1LSRLlge7Ymk9Ru5PsEPMUZdH/XLhczSOzsdSrnkDa4nSAdST5Gf7ju
pWKKDNfeEMuiF1nA1nraV7jHU5twUFITSsP2jJm91BllhbBNjjnlCGa9kZxtpqsO
T80Ow/ZEhoLXt6kDD6+2AAqp7eRhVCS4pnDCqayz0r0GPW13X3DprmMpS1bY4FWu
fJZxokpG99kb6J2Ldw6V90Cynufq3evnWpEbZfCkCl8K3xjEwrKqxHQWhxiWyDEv
opHxpV/Q7Vk5VsHZozBdDXSIqawM/HVGPObLCoHMbhIKTUN9qKMYPlP/d8XTTZfi
1nyWI247coxlmKzyq9/3tJkRaCQ/Aw==
=Wmam<
-----END PGP SIGNATURE-----

If you were so inclined, you could easily find my corresponding public key online and use it to verify that signature.  What would that tell you?

Well, it would say that I have definitely asserted something about the above text: in this case, I’m asserting that I wrote it.  It wouldn’t tell you whether that was true, but it would tell you two things:

  • It was definitely me making the assertion, because nobody else could produce that signature.  This is partly because nobody else has access to my private key file, and even if they did, using it also requires a password that only I know. So they couldn’t  produce that signature without me. It’s way, way harder than faking my handwritten signature.

  • I definitely had access to that bit of text when I did so, because the signature is generated from it. This is another big improvement on a handwritten signature: if I sign page 6 of a contract and you then go and attach that signature page to a completely new set of pages 1-5, who is to know? Here, the signature is tied to the thing it’s signing.

Now, I could take any bit of text that ChatGPT (or William Shakespeare) had written and sign it too, so this doesn’t actually prove that I wrote it.  

But the key thing is that you can’t do it the other way around: somebody using an AI system could produce a blog post, or a video or audio file which claims to be created by me, but they could never assert that convincingly using a digital signature without my cooperation.  And I wouldn’t sign it. (Unless it was really good, of course.)

Gordon Brander goes into this idea in more detail in a post entitled “LLMs break the internet. Signing everything fixes it.”   The gist is that if I always signed all of my blog posts, then you could at least treat with suspicion anything that claimed to be by me but wasn’t signed.  And that soon, we’ll need to do this in order to separate human-generated content from machine-generated.

A tipping point?

This digital signature technology has been around for decades, and is the behind-the-scenes core of many technologies we all use.  But it’s never been widely, consciously adopted by ordinary computer users.  Enthusiasts have been using it to sign their email messages since the last millennium… but I know few people who do that, outside the confines of security research groups and similar organisations.  For most of us, the tools introduce just a little bit too much friction for the perceived benefits.

But digital identities are quickly becoming more widespread: Estonia has long been way ahead of the curve on this, and other countries are following along.  State-wide public key directories may eventually take us to the point where it becomes a matter of course for us automatically to sign everything we create or approve.

At which point, perhaps I’ll be able to confound those of my friends and colleagues who, according to ChatGPT, keep wanting to pinch the credit for my blog.

 

 

 

 

 

 

 

Tweetback

For all its problems over the years, one thing that Twitter has always done quite well is to give you access to backups of your data. You can request a dump of your archive and (a day or two later) download a ZIP file containing all of your Tweets and lots of other data besides, which you could process with your own code, but can also view using an HTML page included in the archive.

As the future of Twitter gets ever more uncertain, I ask for backups more frequently! You might want to do the same… one day you may not be able to.

I had an account in the very early days, and then deleted it, before joining again a little while later. So the first tweet on my current account, just over 15 years ago, says:

Just re-established a Twitter account, with some trepidation

I didn’t have an iPhone back then – it existed, but I waited for the 3G version, which arrived a few months later. So quite a few of my early Tweets are about the fun things I could do with my iPod Touch, though I guess most of them would actually have been submitted using SMS text messaging, as you did back then.

I was also amused to see that one of my earliest messages was while barbecuing in the garden. I was recommending a local brand of sausage, which I can still recommend as one of my favourites: “Musk’s“.

That name, but attached to another owner, was to pop up on Twitter occasionally in the future.

I had actually met Elon Musk a couple of years before, but little did I know that his name would one day be so closely associated with this new social network. Back then, he was mostly known for PayPal and SpaceX, though he did have an intriguing side-interest in electric cars. It didn’t look likely that I (or most of my friends) would ever be able to afford one, though, so his other activities were of more interest.

These days, I still consume that fine brand of Newmarket sausages with the same frequency and enthusiasm as I did back then. Other things have changed, however. I drive an electric car every day, and have barely been in a petrol station for 7+ years. Twitter, on the other hand, I only look at a couple of times per week. Elon Musk is primarily responsible for both of those changes!

Perhaps Methuselah didn’t live 900 years after all?

On Sunday I had a video call with an old family friend, Marjorie, who has just celebrated her 103rd birthday, and is doing well. To put that in context, she was a teenager when the king acceded to the throne. No, not the current king. His grandfather. You know, the one whose daughter reigned for three score years and ten after him.

So I had longevity in mind when I saw Charles Arthur’s link to a rather nice study by S. J. Newman from the Australian National University, looking at the data about other people who have lived beyond 100 years.  The records of supercentenarians tend to cluster in particular geographic areas, and many reasons have been suggested for this.

As the paper says,

The observation of individuals attaining remarkable ages, and their concentration into geographic  sub-regions or ‘blue zones’, has generated considerable scientific interest. Proposed drivers of remarkable longevity include high vegetable intake, strong social connections, and genetic markers. Here, we reveal new predictors of remarkable longevity and ‘supercentenarian’ status.

It’s a nice read, and a good lesson in why those involved in data analysis sometimes need to dig a bit deeper.

The quick summary is that areas reporting large numbers of supercentenarians not only have high degrees of poverty, social interaction, and high vegetable intake.  They are also areas where reliable record-keeping was only recently introduced, or where there may be other reasons for assuming that reports of very old people are not entirely accurate.

Some of this can be deduced by looking at the data directly, for example the fact that “supercentenarian birthdates are concentrated on the first of the month and days divisible by five” or that some areas with numerous residents in their 100s seem to have surprisingly few residents in their 90s.

Sometimes there are other reasons for doubting the reliablility of the data:

The large-scale US bombing and invasion of Okinawa involved the destruction of entire cities and towns, obliterating around 90% of the Koseki birth and death records with almost universal losses outside of Miyako and the Yaeyama archipelago. Post-war Okinawans subsequently requested replacement documents, using dates recalled from memory in different calendars, from a US-led military government that largely spoke no Japanese.

Sometimes, past researchers have had to use their own judgement when assessing individual cases:

Individual case studies often highlight the role of personal judgement, and the potential for both conscious and unconscious bias, during age validation. For example, Jiroemon Kimura, the world’s oldest man, is widely considered to be a valid supercentenarian case. However, Kimura has at least three wedding dates to the same wife, has three dates of graduation from the same school, was conscripted to the same military three times in four years despite the mandatory conscription period being three years long, and has at least two birthdays.

Then there are the financial issues to consider:

In 1997 Italy discovered it was paying 30,000 pensions to dead people

and

A subsequent 2012 investigation by the Greek labor ministry was triggered by the “unusually high number of 9,000 Greek centenarians drawing old-age benefits”, a notable figure given the 2011 Greek census found only 2,488 living centenarians.

Yes, the sad conclusion is that:

Like the ‘blue zone’ islands of Sardinia and Ikaria, Okinawa represents deprived regions of rich, high-welfare states. These regions may have higher social connections, and arguably may have had higher vegetable intakes in the past, but they also rank amongst the least educated, poorest, highest-crime and least healthy regions of their respective countries.

Lastly, you know those reports that the longest-lived are often people whose lives are characterised by smoking, drinking and unhealthy lifestyles?  The paper has some wry comments about that too:

It is unclear why clinically excessive drinkers or daily smokers should survive at equal or higher rates, and increase in population frequency at extreme ages, unless these lifestyle factors are positively correlated with committing fraud or having an incorrect age.

And, referring again to the decisions of the researchers who build the databases:

A reliance on this type of opinion, where qualitative judgements are employed to shape public perceptions of authenticity, seems to be widely considered satisfactory. This seems particularly the case when explaining the otherwise anomalous health habits of supercentenarians. For example, Maier et al. issued a contradictory statement that Jeanne Calment smoked both one and two cigarettes a day for an entire century, followed by the justification that this counter-indication of health could be explained because she “possibly did not inhale at all”.

It was likewise observed that, from age 20 to age 117, the then-oldest man in the world, Christian Mortensen, smoked “mainly a pipe and later on cigars, but almost never cigarettes… he had also chewed tobacco…but never inhaled”.

Why two people would voluntarily choose to smoke for an accumulated 190 years, yet never inhale, was never explained.

Zappi Days

When I installed my home solar system, I also replaced my perfectly-functional car charger with a new one: a Myenergi Zappi. Why?

The Zappi is a popular charger, designed in the UK, and rapidly finding favour in other parts of the world. Here, I talk about what it can do, things you might need to take into account if connecting to a car like the Tesla, and a little bit of magic geekery I set up to make it fit my needs even better.

(Direct Link)

You are my sunshine…

We’ve just had a solar/battery system installed, and because I’ve spent a lot of time planning and thinking about this over the last year and a half, and because it has some slightly unusual aspects, I decided to create a video about it.

This starts off as a pretty basic introduction of solar power, and ends up going into some detail about why mine is arranged the way it is. This means that it’s longish… but I hope it might be of interest to anyone considering installing, or expanding, a similar system.

(Direct link)

Clippy comes of age?

I’m old enough that I can remember going into London to see the early launch demos of Microsoft Word for Windows.  I was the computer officer for my Cambridge college at the time, and, up to that point, everyone I was helping used Word for DOS, or the (arguably superior) WordPerfect.

These first GUI-enabled versions of Word were rather good, but the features quickly piled on: more and more buttons, toolbars, ribbons, bells and whistles to persuade you, on a regular basis, to splash out on the next version, unwrap its shrink-wrapped carton, and install it by feeding an ever-increasing number of floppy disks into your machine.  

ClippyAnd so for some of us, the trick became learning how to turn off and hide as many of these features as possible, partly to avoid confusing and overwhelming users, and partly just to get on with the actual business of creating content, for which we were supposed to be using the machines in the first place.  One feature which became the iconic symbol of unwanted bloatware was ‘Clippy’ (officially the Office Assistant), which was cute for about five minutes and then just annoying. For everybody. We soon found the ‘off’ switch for that one!

These days, I very seldom use any Microsoft software (other than their truly excellent free code editor, VSCode, with which I earn my living), so I certainly haven’t sat through any demos of their Office software since… well, not since a previous millennium.

But today, since it no longer involves catching a train into London, I did spend ten minutes viewing their demo of ‘Microsoft 365 Copilot’ — think Clippy endowed with the facilities of ChatGPT — and I recommend you do too, while remembering that, as with Clippy, the reality will almost certainly not live up to the promise!

Still, it’s an impressive demo (though somewhat disturbing in parts) and though, like me, you may dismiss this as something you’d never actually use, it’s important to know that it’s out there, and that it will be used by others.

 

 

ChatGPT is famous for producing impressively readable prose which often conceals fundamental factual errors.  Now, that prose will be beautifully formatted, accompanied by graphs and photos, and therefore perhaps even more likely to catch people unawares if it contains mistakes.  

The text produced by these systems is often, it must be said, much better than many of the things that arrive in my inbox, and that will have some advantages.  One challenge I foresee, though, is the increasing difficulty in filtering out scams and spams, which often fail at the first hurdle due to grammatical and spelling errors that no reputable organisation would make.  What happens when the scammers have the tools to make their devious schemes grammatically correct and beautiful too?

I would also be interested to know how much of one’s text, business data etc is uploaded to the cloud as part of this process?  I know that most people don’t care too much about that — witness the number of GMail users oblivious to the fact that Google can read absolutely everything and use it to advertise to them and their friends — but in some professions (legal, medical, military?), and in some regimes, there may be a need for caution.

But it’s easy to dwell on the negatives, and it’s not hard to find lots of situations where LLMs could be genuinely beneficial for people learning new languages; struggling with dyslexia or other disabilities; or just having to type or dictate on a small device a message that needs to appear more professional at the other end.

In other words, it can — to quote the announcement on Microsoft’s blog page — help everyone to ‘Uplevel skills‘.  

Good grief.  Perhaps there’s something to be said for letting the machines write the text, after all.

Back again… for the anniversary

Status-Q, in its current form, is 22 years old today. I haven’t posted for a week or so because I’ve been terribly energetic on some Swiss ski slopes.

My legs wearing ski boots, reclining on a deck chair, with snowy swiss mountains in the distance.

A return to “Ulysses”

After my rather damning post about James Joyce’s Ulysses a little while ago, I decided I needed to make a more determined effort to sample more of it so that I knew whereof I spoke. I would make a serious attempt upon its slopes.

A couple of people suggested that it is better to listen to it being read by a good Irish actor than to read it yourself, and that made sense to me. Many people prefer Shakespeare when acted… but of course Shakespeare was meant to be acted. It’s also written in the language of four centuries ago. Ulysses is modern and was meant to be read, but it helps if a good actor interprets it for you because Joyce was either too lazy or too pretentious to make it easy for you with things like normal punctuation, or, in many cases, identifying who is actually speaking.  

I thought Hilary Mantel’s use of the third person present tense in the Wolf Hall trilogy was also a somewhat foolish affectation, but it turns out to work exceedingly well in audiobook form, giving it the immediacy of a screenplay. It’s a much more enjoyable read than Ulysses, but, again, it helps to have somebody interpret simple things like who the author actually means by ‘he’ in this sentence! It’s not that readers can’t work it out, it’s just that the need to do so imposes friction which is unnecessary and doesn’t, to my mind, add anything. Mantel was a good writer and didn’t need to resort to such novelties to get attention.

Anyway, I love audiobooks, and spent my one Audible credit on this reading of Ulysses by Tadhg Hynes. It runs to nearly 32 hours — almost a French working week — and both he, and Kayleigh Payne, who reads the part of Molly, do a really terrific job. So with their aid, I have now ‘read’ the majority of Ulysses; at least, I got well past the 50% mark, and then skipped to the last chapter. But even having read about two-thirds of it gives me a sense of achievement rather like having finished an exam or run a marathon. I achieved something, but I don’t have any desire to do it again.  Perhaps, since I skipped ahead when the tedium became too much, I can at least say I managed a half-marathon!

I was trying to work out, as I wandered the country lanes listening to it, what attracted people to Ulysses. It has originality, certainly, on various fronts. Few novelists spend as much time discussing defecation, masturbation and menstruation, for example, and one could argue that they therefore miss out on key parts of the human experience! So it’s certainly memorable in places.

But overall, here’s my Ulysses FAQ:

  • Is it compelling? No, it’s a long slog, requiring serious stamina.

  • Is it entertaining? Only rarely. I would chuckle lightly once every few hours.

  • Does it have a good plot? No. It has no discernible plot.

  • Are the characters interesting? No.

  • Is it edifying? No.

  • Is it educational? No.

  • Is it beautiful? No.

  • Is it clever?

Ah, well, there you have me. Yes, I have to admit that it is often very clever. And there’s the rub. I imagine that the more you study it, the more you see in it, and the more you appreciate the cleverness. People with lots of time on their hands — those taking Arts degrees, for example — can probably unwrap many more layers than I did, and so are more likely to think it great.  If you’re an English Lit. student and can set aside a week to listen to it while also reading it, and pause periodically to consult study notes explaining it, then the book would, I’m sure, yield a lot more.  Joyce’s intention was to make it obscure, saying that he had “put in so many enigmas and puzzles that it will keep the professors busy for centuries arguing over what I meant, and that’s the only way of insuring one’s immortality”. In that, he succeeded! I’m sure it’s great material for literary PhD theses.  

But here’s the thing… Is it actually worth that effort? It seems strange that an author should go to so much trouble just to be clever.  I spent about 20 hours in the company of the main characters, for example, without developing the slightest interest in them or any desire to go any further in their company. It’s almost as if the book deliberately has no story, no purpose, except to be clever. And that, I think, is its chief failing.

Now, there is some undeniably great poetry in there. Occasionally Joyce would capture in four words what might have taken others twenty-four, and I would tingle with appreciation. But there were far more occurrences of using twenty-four words for what could have been said in four, or probably eliminated completely, if he’d had a better editor.  (The fact that I felt able to skip the equivalent of about 200 pages without much risk of missing anything important will give you the idea.)

To return to my marathon analogy, it’s as if you’ve been condemned to run for hours through dull industrial estates and suburbs, so when you see the occasional cherry blossom you gasp, “Oh, how beautiful!”.  The real question is “Why didn’t the organisers pick a route through beautiful countryside in the first place?” The book has interesting aspects, but it could have been so, so much better if it had had a plot, interesting characters, less pretentious forms of originality, less unnecessary friction. A writer with the capabilities that Joyce very clearly has could have chosen to overlay his brilliance on a much better landscape, and then he would have written a truly great book.

So I think my conclusion is this: Ulysses has the lowest signal-to-noise ratio of anything I’ve ever read or listened to. It’s like straining hard to detect the snatches of music through a radio with too much static, and discovering in the end that the tune just wasn’t very good.

Gasoline car review

A wonderful review by Geoff Greer compares one of these gasoline-powered cars to the experience of driving a normal electric one.

Here’s an excerpt:

Refueling was a mixed bag. Whenever I parked in my garage, I had to suppress the urge to plug the car in. Gasoline is only available at special stations, and it is prohibitively expensive to get a gasoline line installed in your home. So unlike a normal car, you don’t wake up every morning with full range. The only way to add range is to go to gas stations. These are similar to fast charging stations, but smelly and more dangerous.

I’ve been driving electric cars for over seven years now, so the very rare occasions when I visit a petrol station — a couple of months ago to buy some screen wash liquid, for example — definitely do remind me of what I’m (not) missing!

Anyway, do read the whole thing.  

Thanks to John for the link.

Infinite dominoes

This is a lovely bit of design. I’m not sure whether I’m more envious of someone who has so much Lego, or someone who has so much time to play with Lego…

(Direct link)

Language Skills for Time Travellers

Greetings, O time-traveller from the past!  Welcome to our decade!

You may have noticed some distinct eccentricities in the English spoken in our time, but here is a quick phrase book to help you understand some of our more curious modes of speech:

2020s English Traditional English
“What’s happenin’?” = Good morning/afternoon/evening
“upskilling” = training
“to incentivize” = to give an incentive to
“cosplay” = dressing up
“going forward” = in the future
“reaching out to” = contacting
“he gifted” = he gave
“wellness” = feeling happy
“poor mental health” = feeling unhappy
“tweeting” = (actually, you probably don’t need to bother about this one any more)

Now, let’s put what we’ve learned into practice. Here’s an example message that might be received by an employee of a company in the early 2020s from the senior management. Using the above table, you should be able to understand it:

What’s happening, everybody? Great news! We’re reaching out as part of our wellness and incentivization program to let you know that, going forward, we’ll be gifting you everything you need to participate in the cosplay sessions, which will take place on the evening before each of the monthly upskilling days. Remember, regular social activity has been proven to enhance your mental health!

What? Wait a minute…! You’re not staying?

Q Tips

Some simple tricks for Mac users.  Do you know all of these?

 

Direct link

© Copyright Quentin Stafford-Fraser