Edition 4

My Dearest Reader,

I write to you by the scant flickering stub of our last tallow candle. Alas of the sixty-four who set out on this ill-fated expedition only two of us survived. This evening we faced the terrible burden of choosing between using said candle for illumination or for sustenance. In the end I opted for the former and resolved the conundrum by consuming my companion. In an irony I’m sure he’d have appreciated as my former gamekeeper he was actually rather gamey…

Sorry about that, obviously total immersion in the world of AI is driving me to the edge of bonkers, but then becoming slightly unhinged seems to the sanest strategy in the face of something so big and daunting it could probably beat an abyss in a staring into contest.

Terry Pratchett once observed that ‘war is a crime so big you can’t see it’, in a similar sense AI is so colossally huge it is almost impossible to grasp the enormity of it.

But there is method in my you know what. Those who’ve taken on any large task will know the secret is to break it down into manageable chunks. This is why I’ve taken to using (overusing) the ‘AI is Rome’ metaphor, it’s quick, dirty, probably not that accurate but accurate enough for now. In the last couple of weeks, I’ve even narrowed it down to a more specific time of the empire’s history

I think we are in the Trajan era. Why? Well, Trajan was groundbreaking in being the first non-Roman Roman. He came from the province of Hispania Baetica (modern day Andalusia) and until him only Romans born in Rome were considered suitable to rule. Trajan’s ascent to Emperor heralded the idea that Rome was no longer about Rome- it had (like AI and Pratchett’s idea of War) become so big it no longer fitted into itself.

The sheer unprecedented scale, size, and expansion of AI means it to has now burst far beyond its original boundaries and is quickly seeing the world as its Mare Nostrum.

However, despite our natural inclination to see Rome as the original power mad Fascists its expansion was never actually planned, proof for that is to look at what an architectural gumbo the ancient city of Rome was – all those geometric roads and precise civic planning was for everyone else but not for Rome itself - believe it or not at one point the Eternal City had four streets that actually crossed themselves!

I think AI is similarly expanding not through intent but because that’s ultimately what systems with optimised self reinforcing positive feedback loops always do, be it empires or civil service departments, social media sites or religions – as soon as they find a ‘spinal equilibrium sweet spot (strong enough to be firm like a solid yet at the same time flexible enough to act as a fluid – when this occurs the system expands as a result of the natural dynamic push pull between the two exactly what Humans did the moment we stood upright and ‘grew a backbone’ and now AI appears to be following the same path.

Anyhow, that’s enough about Romans for now, just pass me those olives, apparently the next bout features one of them weird blokes with the net and trident…

A few AI Updates

Adobe, Nvidia, Palantir, OpenAI, Amazon, Anthropic, Google, Inflection, Meta, Microsoft, and Stability AI have now all signed up to White House's AI pledge commiting to AI safety

Interesting piece on how AI and the threat to Journalitic integrity

Coca Cola obvioulsy feeling miffed and FOMO

A Prompt Prompt Apology

A quick retraction of last issues prompt declaration: My client (me) wishes to state he (me) is not nor never has been anti-prompt and he (I) regret that anyone may have interpreted his (my) comments as being perceived as any kind of attack on the prompt community. My client (still me) has always been a strong defender and champion of Prompting.

What I meant was that prompts are great but are only as great as the person creatively wielding the prompt – as the ancients saying goes ‘a book is not a mirror; if a monkey looks in a wise man will not stare back.’

The Ever-Ensnaring Enigma of AI

The realm of AI is undeniably strange. This sentiment is captured interestingly by the linguistic relationship between the words "strange", "stranger", "strangle", and "struggle". All these terms trace their origins back to a root meaning that evokes a sensation of choking of being constricted. This illustrates one of the challenges of trying to grasp the concept of AI without thinking it is a bit strange. To put it simply AI is just not anything of the things we naturally want to pigeonhole it as, its not a glorified thesaurus or a clever search engine or a database or a vast online library. It is all of these and a myriad of others all at the same time And that’s the problem; the more we attempt to label it as something the more constrictive it feels and lo, we’re back to the feeling of being uncomfortably restricted.

The innovations and advances and discoveries in AI keep coming thick and fast. To cope I’ve tried to remember lessons from Claude Shannon the Pioneer of Information Theory and worthy of the title of Grandfather of AI. Shannon’s brilliant ideas can be seen as a precursor to the age of Artificial Intelligence.

At its core, Information Theory posits that the essence or amount of information within a message is directly tied to the level of surprise or novelty it presents.

For instance, a statement as short as "Aliens have just landed!", the phrase ‘oh-oh'‘ or the equation E=mc2 carry vast amounts of information due to the implications / amount of surprise. On the contrary, a long sequence like "111111111111..." conveys minimal information, however long it is.

In today's age, we’re bombarded with claims of AI achieving astonishing feats. However, the frequency of these announcements has now begun to make them less surprising over time.

In such an environment, there's a danger of constantly chasing the next big thing. Especially when the landscape is stuffed full of so many "shiny" innovations, it risks intellectual fatigue. It’s even got its own name ‘The Midas Problem’ – what do you do when everything you touch upon comes off as rich, inviting, and lovely? (all good until you have to pee!)

To provide some perspective, just this morning, I reviewed my compiled list of active AI apps and services. The spreadsheet runs to over 1,847 separate entries, a testament to the overwhelming advancements in the field but also to the absolute gold rush feeling of it all. With that many entries how the hell can you possibly test and review? Another compilation I have is all the best advice on improving your results when using AI, that one now stands at just under 5,000 entries

I’m now at a point where I’m using AI to assess the efficacy of itself. Which is like handing the exam paper to the person sitting the exam and saying, ‘can you test this test and let me know if the test is a good enough test for you to pass it?

You can see why I said becoming a bit unhinged is a viable strategy - in the meantime to help AI&I (that’s me and AI working together) came up with this as something to watch out for when it comes to all things AI

Overly Curated Narratives:

Definition: An "overly curated narrative" refers to a story or account that has been meticulously and selectively constructed to present a specific perspective, often omitting certain details, challenges, or setbacks to create an idealized or favorable image. It implies a lack of comprehensive or balanced information, potentially leading to a skewed or biased understanding of the topic at hand.

Usefulness:

1. Critical Analysis: Recognizing an overly curated narrative allows readers or listeners to approach information with a discerning eye, prompting them to seek additional or alternative sources for a more holistic understanding.

2. Media Literacy: In an era of information overload, being aware of such narratives aids in media literacy, helping individuals differentiate between genuine, balanced reporting and content that might have ulterior motives or biases.

3. Credibility Assessment: By identifying gaps or selective presentation in stories, one can assess the credibility of sources and decide how much weight to give to their claims.

4. AI and Tech Scrutiny: As you pointed out, in the context of AI and technology, where claims can sometimes be grandiose, the concept is particularly valuable. It encourages stakeholders, investors, and the general public to question and probe deeper into claims about capabilities, impacts, and ethical considerations.

5. Personal Growth: On a personal level, being wary of overly curated narratives can prevent individuals from comparing their own lives or businesses to idealized versions presented by others, leading to healthier self-assessment and growth.

In essence, the term "overly curated narrative" serves as a reminder in our information-rich age to approach stories, claims, and reports critically, seeking depth, authenticity, and balance. It encourages a comprehensive understanding rather than accepting information at face value.

A final bit of advice I need to take more of myself is to embrace a Buddhish approach in understanding that everything is in a permanent state of flux and that change really is the only constant.

How the other half rule

Thought it was interesting to see how Jensen Huang runs Nvidia a company estimated to be worth 1.12 trillion dollars (yes, you read that right)

40 direct reports, no One to Ones

• Believes the flattest org is the most empowering one, that starts with the top layer

• Does not conduct 1:1s everything happens in a group setting

• Does not give career advice "None of my management team is coming to me for career advice - they already made it, they're doing great"

No status reports, instead he "stochastically samples the system"

• Doesn't use status updates because he believes they are too refined by the time they get to him. They are not ground truth anymore. - Instead, anyone in the company can email him their "top five things" with whatever is top of mind, and he will read it - Estimates he reads 100 of these every morning

Everyone has all the context, all the time

• No meetings with just VPs or just Directors - anyone can join and contribute

• "If you have a strategic direction, why tell just one person?"

• "If there is something I don't like, I just say it publicly, I do a lot of reasoning out loud"

No formal planning cycles

• No 5-year plan, no 1-year plan

• Always re-evaluating based on changing business and market conditions (helpful when AI is developing at the pace that it is)

• This org is optimised for (1) attracting amazing people, (2) keeping the team as small as it can be, and (3) allowing information to travel as quickly as possible

This does seem much more akin to leadership will be best suited to be in an AI world.

And now finally, as so often promised

The Gorilla Problem

About a few million years ago, our ancestors and gorillas diverged from a common lineage. This divergence led to two distinct evolutionary paths: one resulting in modern humans and the other in gorillas. While gorillas have largely remained unchanged, we humans have evolved significantly. At one point, both species were on equal footing, but today, gorillas exist primarily because we have decided to preserve them. We've carved out habitats for them where they can live peacefully. Whether they are truly unaware of this or simply remain silent about it, is a mystery.

One thus must ponder: As we advance in the field of AI will we eventually find ourselves in the same position relative to AI as gorillas are to us?

My own slight tangent is you must consider the nature of power. Power and dominance are often linked to the instinctual drive to procreate. The question then arises: Will AI ever develop such drives?

This situation is reminiscent of theories suggesting the existence of advanced alien civilizations capable of interstellar travel might invade Earth, akin to a 12th-century Viking raider.

If a species possesses the intellectual capacity for space travel, what are the odds that they still retain primitive raiding instincts? It's plausible, but you’d have to question the logic or reasoning behind such actions. I think the same goes for AI.

Right, that’s it for this week, be off with you, I’m sure that £20k wrinkle removing cream made from the tears of a mummified ancient Inca won’t apply itself.

Vincimus Minimus