This newsletter is written for entrepreneurial leaders who want to learn about the moment we are living in but don’t have time to read broadly; who want to grasp the key themes; and who want to create better ways of advancing their mission. The Weekly Distillation covers a broad range of topics with the intent to curate the key narratives of the week, how they fit the broader themes of society and to pose questions that help you to think deeper on the application in your context. You can read more about the key themes I see here.
Someone once said
"Orwell's '1984' is more like a general warning than a prediction, for in times of universal deceit, telling the truth becomes a revolutionary act." - Noam Chomsky
"By far, the greatest danger of artificial intelligence is that people conclude too early that they understand it." - Max Tegmark, physicist, and AI researcher
"Writing is easy. All you have to do is cross out the wrong words." - Mark Twain
"If you spend more on coffee than on IT security, then you will be hacked. What's more, you deserve to be hacked." - Richard Clarke
“Mr. President, I feel I have blood on my hands,” - Oppenheimer
What’s artificial about intelligence?
I do love a Gartner hype cycle. We saw it in the late 1990’s (office procurement dot coms, supply chain software), 2000s (search engines), 2010s (social networks), 2020 (crypto), 2021 (NFTs, DAOs), 2022 (errr maybe not), 2023 (LLMs, Psychadelics; AI generally & ChatGPT specifically; Bitcoin on the recovery side of the chart). Last week it was photo AI, this week it was video AI generated from photo AI. I still think Johnny Cash singing ‘Barbie’ is incredible.
Before I go all dystopian on all y’all (sorry, have just been in Florida and listened to too much Luke Combs) here’s a few apps I’ve been bookmarking for you to check out:
SumUp: Make articles easier to understand using AI - https://sumup.page/
Mind Studio: Create personalised, context aware AI apps in minutes - using no-code solutions - https://youai.ai/mindstudio
Mayday: AI assistant + calendar that helps you organize, schedule, and protect your day. - https://mayday.am/
Composer: Build trading algorithms with AI, backtest them, then execute - all in one platform. No coding skills required - composer. trade
Right, that’s enough of that.
What was your vacation reading? I dusted off 1984, a George Orwell classic that I had never actually read. What a great (and very disturbing) read for this current moment. Here’s a frame of how I’m thinking about AI currently and where we go from here.
Level 1: The first stage seems to be the ability to capture data. Back in the 1970s and 1980s most if not all of our technology was communicating one way. Your Sony Walkman for example. (If you were born after that was around, look it up - it was a game changer). Now we go from sonar buoys, to internet of things, smartwatches, Alexa, phones that maybe or maybe don’t listen to you but are certainly capable of it even if switched off, smartTVs, CCTV, satellites, smart buildings, connected cars, drones - you get the point. We have created the ability to collect an unbelievable amount of data.
Level 2: Data is nothing by itself. I’ve worked on a big data project over the last year and without the analysis and cleansing it is all just meaningless noise. The 2000s saw the rise of data cleansing and analysis (Cambridge Analytica anyone) and the ability to provide reports and insights from big data trends. Scraping / Gathering / Extracting and then Transformation / Cleansing, followed by generating semi-structured or structured data. Then the clever stuff in the data mining and analytics, followed by the data scientists garnering insights and using visualisation tools to load it all up into nice dashboards and reports. Again, all good so far, right? Yes, if you have a lot to hide then maybe not - but if not, why worry? Let’s come back to that.
Level 3: Hello to the democratisation of AI databases and the tools that can be built off it. Large Language Models (LLMs) have arrived, enabling Generative AI apps such as ChatGPT. LLMs are trained on trillions of words over time, leveraging incredible computing power to do it.
One interesting side story is the rise of Nvidia. I went for a pie and pint with a friend last night and we got to talking about chips (not the kind you eat, we just ate those ones) and the dynamics of the industry. Back when all my spare time was applied to developing my Football Manager / Championship Manager game skills, and I needed to increase the graphics processing capability of my desktop, an Nvidia GPU was the answer. Somewhere in recent years, Nvidia realised that GPUs are incredibly powerful processors and can be applied to AI calculations. They haven’t looked back and in May were valued at $1 trillion. Best pivot of all time?
ChatGPT tangent - what do you think about that then my wonderful AI assistant? Give me the best 3 pivots of all time.
The term "pivot" is commonly used in the business world to describe a significant change in a company's strategy, product, or focus. When discussing the "best" pivots of all time, it's essential to consider the impact, success, and transformative nature of the changes made. Here are three notable examples of successful business pivots:
PayPal: Originally founded as Confinity in 1998, the company initially focused on creating security software for handheld devices. However, they recognized the potential of their money-transfer service and decided to pivot their business model to focus on online payments. In 2001, the company rebranded as PayPal, providing a secure and convenient platform for online transactions. This pivot proved to be immensely successful, revolutionizing online payments and eventually leading to its acquisition by eBay in 2002. [EDITOR - OK, GOOD ONE]
Nokia: Before becoming a renowned mobile phone manufacturer, Nokia started as a pulp mill and diversified into various industries, including rubber products, electronics, and telecommunications. In the 1980s and 1990s, Nokia realized the potential of the mobile phone market and strategically pivoted its focus to become a dominant player in the mobile industry. This shift enabled Nokia to become one of the leading mobile phone manufacturers globally for many years. [EDITOR - YES, BUT YOU COULD HAVE GONE SONY STARTING ON KETTLES OR ME WHEN I USED TO SELL MERINGUES]
Twitter: Twitter's origins can be traced back to a podcasting company called Odeo. When Apple announced its plans to integrate podcasting into iTunes, it posed a significant threat to Odeo's business model. The company's co-founder, Jack Dorsey, came up with the idea of a microblogging platform that allowed users to share short status updates. Odeo decided to pivot its entire focus to this new idea, and Twitter was born. Today, Twitter is one of the most influential social media platforms globally. [LOL, THAT LAST LINE…..BING AI WOULD AT LEAST BE UP TO DATE POST 2021]
These examples showcase how successful pivots can dramatically transform companies and lead to their long-term success. Pivots often require vision, adaptability, and the willingness to take risks, but when executed effectively, they can lead to game-changing results.
Back to LLMs. The image below explains the current quite clearly (or at least enough to get the point). If you want to go deeper into the specifics of each one then go here.
We can now create (generate) content (video, images, text, answers, essays, tables, code) from these LLMs from the apps we create. I still find descript the most disturbing. Want to sound like someone else? Or swap your singing voice? GPT 4 is a step change from GPT 3 apparently.
Level 4: So we have massive data, captured, cleansed and analysed. We have LLMs trained on hundreds of billions or trillions of parameters. We have apps on top of these being released by the hour. All good and mildly interesting so far. Where do we go from here?
A) The Malign Actor. I changed my views on big data around the time I did a module in cyber security at the Handa Centre for the Study of Terrorism & Political Violence (part of the University of St Andrews) a few years ago. Reading a few books, including Cyber War by Richard Clarke, advisor on cybersecurity to the US President in the early 2000s. It became apparent that it’s too late to worry about protecting your data as c.80% of Americans already have had personally identifiable data stolen by the Chinese alone. Add TikTok to that list. And this is leading to death and geopolitical consequences already.
B) The one that had me in deep reflection on vacation was this. What is to stop the automation of the consequence of the analysis? For example:
X country uses sonar buoys to capture sounds of boats in territorial waters. Satellite and drone images are used to match those to particular vessels, landing points and disembarkation. Data cleansing and analytics identifies that these are smuggler boats. The programme has recognised from its LLM that most people don’t like smugglers and the population (in general) wouldn’t be too sad if a few were sunk to send a lesson. An automated command is sent to the drone (with no human oversight) to fire a Hellfire at the next boat that matches the sounds profile and visual match.
A zealous and insecure senior manager asks an AI app to identify the 10 employees worldwide that gossip most about their bosses on work platforms, and are least likely to be hired elsewhere. The tool matches this with productivity outputs and quickly suggests 10 employees who should be let go. The system automatically sends an email terminating their employment.
A country determines that criticism of a particular policy is unacceptable. For example, the turmoil in Israel about the Supreme Court not being able to overrule the Knesset. AI tools scan social media, websites, other public spaces and identify candidates using ‘wrong’ terminology. Restrictions are placed on their ability to access certain spaces, speak at certain events, access certain public services and even spend in certain places. Or what about identifying criminals or drug dealers from AI?
All hypothetical? The question is not the first three levels (although there are big questions in there too), the big question is whether or not we will have safeguards, ethics bodies, oversight, laws and all the appropriate checks and balances to ensure that development is done thoughtfully and with wisdom to weigh the consequences. I am not optimistic on this. Our track record is to build technology and then regret it later.
Some questions for you:
How do you know the AI app you’re trialling isn’t just another fake tool to capture your data?
Who might suffer from the rolling out of this technology? What impact is that having on your utilisation of the technology?
How do you ensure that what you create from Generative AI adds to the world and isn’t just more noise?
What do you think we should do to leverage the power of AI and protect society from the dangers?
I leave you with this.
Closing thought
I’m back. After not writing The Weekly Distillation since September 2022, I am sure many of you thought this was dead and buried. To be honest, I wondered it myself. A combination of demands of work, and writing here and here about whisky, meant that this Substack had to take a back seat. After some recent time away, I’ve returned with a plan desire to relaunch TWD. Thanks to everyone that emailed or spoke to me to say they loved this newsletter and wondered when it was coming back.
I love that The Weekly Distillation is back! And a great read too. Though difficult not to stick one's head in the sand at the overwhelming-ness (probably not a word) of the whole AI situation. About AI and about climate change, I often feel a sense of panic, wanting to take responsibility for my own part and wanting to know what and how I can do something, but not really coming up with the answers... And if everyone else feels the same, wanting to take action and not really knowing how, what does the future look like?