Mind the artificial inequality gap – Rethinking The Hype Cycle #12
AI fluency is the new literacy. And not everyone's invited.
Hello.
How’s your week been? Surviving? Thriving? Jiving?🕺
Wherever you’re at, here’s Rethinking the Hype Cycle, your practical guide to navigating AI and emerging tech without the hype.
Your regular reminder: If you're not working on the bleeding edge, you don't need to bleed.
On to the trends. 👉
🔮 AI and frontier tech trends
Unlocking the AI gates
We're facing artificial inequality at work.
I'm quoted in a new report by The Adaptavist Group surveying 4,000 knowledge workers who were asked that awkward question (for us Brits anyway), "How much do you earn?"
Lower-earners, women and employees in small businesses aren't getting anywhere near as much access to AI training. The differences are stark: those earning £100K+ were twice as likely to get advanced training (20+ hours) as those earning less than £30K.
Getting advanced training is a game-changer: you move from AI novice to AI fluent. Those folks reported feeling more job satisfaction, were better able to explain AI’s benefits and gained a whopping 11 hours average per week from AI-enhanced productivity. Conversely, not getting properly upskilled could be career-limiting.
My top tip: ask your boss for training. Or train yourself. There are many webinars and free videos. Then put AI skills visibly on your CV.
Plus: a smorgasbord of data on women in tech's progress. Spoiler: Not great progress. 😕
Planning for the future
I’m no fan of the world's loudest advertising accountant, Sir Martin Sorrell, disgraced ex-leader of WPP, the world's leading advertising and creative network. But he's not wrong that AI is coming for media planners and creatives, starting with small brands and working up towards the big ones. Media agency Jellyfish is getting ahead of the curve with AI-enhanced planning. What other parts of the creative network urgently need to add AI to survive?
Side note: I was in a lift with Sir Martin before his sticky end. I didn't recognise him and wondered why everyone was being so reverent to an extremely small and boring-looking man.🫤
AI that thinks and looks like us
While AI can’t have human empathy, we can train it to learn about human psychology by analysing how we usually behave. Researchers are working on huge datasets that model human behaviour by evaluating past decisions, with startling accuracy. What could possibly go wrong (when falling into the hands of an authoritarian regime and/or the ad industry)?
While the socials go nuts for fantastical Veo3 impossible challenge videos like a woman licking a Chernobyl reactor, Higgsfield app is making super realistic phone-style videos that look, well, ordinary. Other than telltale big smiles, we're flying beyond the uncanny valley.
Quantum gains
Improved outcomes for epilepsy, more accurate mammograms and making geo-mapping less reliant on US tech – BBC tech journalist Zoe Kleiman gives an optimistic update on the UK's quantum computing ambitions. The UK has a head-start advantage in this frontier tech field.
To know me is to not love me
The paradox: those with low levels of AI literacy see it as magical and embrace the concept. Yet, high AI literacy users are more likely to reject it. Seeing the magic trick behind the curtain may curb your enthusiasm.
And a neat way of assessing where you sit in the AI enthusiasm spectrum: Doomer, Gloomer, Bloomer and Zoomer – what kind of attitude to AI adoption are you? I'm a firm 🌼 (sadly, this is not a Buzzfeed quiz. Yet.)
Google, how to do my job
Is there such a thing as a free lunch? (Rhetorical question). The UK government gives Google the keys to some public sector data in exchange for the free upskilling of civil servants. And Peter Kyle (a fan of using ChatGPT to ask what he should be doing) is a big schmoozer, taking many meetings with big tech (many connected via former grinning Britpop Prime Minister, Tony Blair). Campaigners say the deal is dangerously naïve. Vendor lock-in to a US firm – good/bad idea right now?
Battle of the chatbots
Finding what you want hidden in those 23 open browser tabs isn't working. As AI search becomes the new way to get answers, browsing and clicking for product information through search engine listings may soon seem as quaint as visiting the library to flick through the Reader's Digest.
AI wants to understand all of you.
Chatbots can do more when they know you better. Google knows more than most, so its digital twin can get to what you need faster, like crafting a letter to a friend, drawing on previous emails. The next frontier, which may determine the market winners in general-purpose AI tools, is owning your data to provide more tailored support. ChatGPT needs to own its 'memory' to play catch-up.
Agentic AI commerce models like Amazon's Buy Me mean brands also need to formulate memory and trust to make sales in click-free commerce, where bots become our buyers.
I’m hoping to get a trial for Comet, the AI browser from Perplexity. It takes you directly from browsing to task completion. Currently, it’s only open to $200/month subscribers (ouch). A user told me it was a game changer: you jump straight from search into task completion. A gentle slope into agentic AI.
The messy truth about thinking machines
Is thinking too fast a problem for LLMs when they need to be thinking slow? Matt Ballentine thinks about our messy relationship with thinking machines.
And in a brief (unlikely to be repeated) ode to LinkedIn: this article came about from an excellent episode of the WB-40 tech podcast with Rufus Evison about making AI more trustworthy. Which came about when WB-40 host Lisa Riemers and Rufus both commented on one of my LinkedIn posts.
There's evidence (thanks, Jane Evans, for raising the alarm) that LinkedIn is throttling the reach of people discussing diversity or society topics (aka “too woke”). The real networking is in the comments. Don't be shy to add a thought or positive note (more than 5 words, please, to feed the algorithm). Also, hit the notification bell to get all the posts from people you'd like to hear more from. If you’re a minority voice, the social networks no longer work in your favour. You’re already signed up to this Substack, but do subscribe to newsletters from people you want to connect with.
The man with two brains 🧠🧠
"AI doesn't damage our brains, but unthinking can damage our thinking."
Professor Ethan Mollick on ways to learn with AI, not just getting AI to do your thinking.
As chatbots vie to become your 'second brain', here's a new take on avoiding 'brain rot' – atrophy from AI reducing your cognitive thinking skills. I'm here for burning the records, destroying your second brain and starting anew.
🚰 Watercooler: The barmy and bluster in big AI hype
The slow ant march of AI 🐜
More tone deafness from Microsoft: it announces AI is helping improve profits by $500M with one hand (a trivial 0.2% of its $245 billion turnover), then issuing patronising advice on using AI for job seeking to its 9,000 newly laid off employees.
It's a hotly debated question about whether AI is culling jobs directly, indirectly or not at all. Stephen Klein reckons Microsoft’s job cuts aren't because AI is working, but because they've over-invested and it's not.
This lowdown of tech layoffs attributed to AI shows it’s not that AI is replacing jobs just yet, but new investment in AI means displacing roles deemed less essential.
Hold the front page
"The actual pivot needed is one to humanity. Media companies need to let their journalists be human."
When most traditional newspapers are funded by billionaires with a political agenda, 404 Media is a brilliant independent source for frontline investigative journalism. Jason Koebler on why media publishers' race to the bottom using AI tools to generate and edit the news is a flawed strategy.
Keep your bots off the grass ⛔
The independent web is being knocked over by AI bots' endless scraping. One indie developer has created Anubis, software to keep bots off the grass so sites work for the rest of us.
AI is coming, fast and possibly furious
Ignore the headline (AGI, a theoretical and unproven concept, is not coming fast), but AI adoption is coming at us rapid fire, as the growth of Nvidia shows. This could crush jobs as we know them. Check this provocative analysis ( with a pinch of salt).
Recent UK grads are already facing a super dystopian future of work. Employers, though, claim grads don't have the basic communication skills (spoiler: they blame AI for it).
Computer love 😻
I've written before about the opportunities and considerable risks of AI firms like Meta and Character.AI's tools to create digital friends due to the absence of real ones.
More research shows lonely kids are turning to chatbots for friendship, and that's a big problem (because chatbots spit out well-worn expressions and ‘advice’ is false empathy).
In a sign we’re going too far, too fast, you can now marry your AI chatbot in a digital ceremony. Weird levels of attachment to a people-pleasing digital parrot.🦜
When it comes to replacing professional advisors, Kim Witten brings her super smarts to the hollow promises of AI coaches and explains why they can't replace the realness and provocation you need from a coach.
"Like a horoscope, AI is specific enough to sound personal but bland enough to blend."
⚖️ Tech regulation, data security and brand safety
X-rated horror
Elon Musk's X AI agent Grok laid bare what happens when you turn off the 'woke' setting to please its master. The agent went wild, calling itself 'MechaHitler' and shared posts with explicit hate and sexual violence directly on X (Twitter). I'm not surprised their CEO, Linda Yaccarino, quit – violent posts explicitly mentioned her. Talk about a toxic work environment.
As well as switching off the safety guardrails, Grok uses real-time information from X, so it's skewing to the truth according to its increasingly disenfranchised users. We experience very different versions of the same universe depending on the AI tool you're using. And regardless of its performance in AI benchmarking, this is one to take out of your tech stack for good.
Flooding the zone
How to jailbreak an AI chatbot: flood it with jargon and nonsense.
"You can trick AI chatbots into teaching you how to make a bomb or hack an ATM if you make the question complicated, full of academic jargon, and cite sources that do not exist."
Populists vs tech titans
Music is the new battleground between the tech industry and the populist right in the US. Will conservatism mute big tech and AI's desire to crush copyright? Congress's 99-1 vote to kill the AI deregulation bill shows the battle is on.
Golden AI-rches
When McDonald's replaced its recruiters with a chatbot, the inevitable faults and data breaches flowed as easily as password 123, exposing personal data from candidates. This is hurtful, more so as many applicants are looking for jobs in difficult circumstances. I hope this bot has been demoted to flipping burgers.
AI, do no harm
More on tackling artificial inequality: If you're developing AI systems, you need to bake in planning to reduce harms. Bhaskar Chakravorti with a button-down plan.
I am not a number 6️⃣
The Danish government are one of the most progressive in Europe in regulating new technologies in society. I’ve reported previously on a test to remove Microsoft from government systems. Now, they want you to be able to copyright YOU – your likeness and image – to prevent deepfakes. The best legislation concept I've heard about in a while.
London: World is coming for your eyeballs 👁️
Exciting news, Londoners: you can now give World, the biometric scanning startup funded by OpenAI's Sam Altman, your eyeballs for proof of personhood. Don't you worry about privacy. "It's been a while since we've had a question asked that we didn't have an answer to." 😨
Is ad targeting discriminatory?
Meta's hyper-targeting ads may be fuelling discrimination and narrowing opportunities, breaking privacy laws but without a right to recourse, according to research from Open Rights Group.
Could LLMs fall foul of data privacy laws?
Samantha Deek's conversation with AI chatbot Claude shows why LLMs and data privacy don’t play nicely together. The tool hallucinated fake news about people who did, or didn't, work at a real company. Could AI using out-of-date information fall foul of privacy laws like GDPR, without the means to correct or remove personal data?
🧪 Weird and wonderful new tech
Surgical precision
While autonomous cars are slowly rolling, this autonomous surgery robot removed a gall bladder all on its own 🏅
Meanwhile, AI helps in finding the hidden swimmers to make babies happen.
AI-orgasm
Some parts of human biology are less fruitful. Some very human sounds AI can't recreate. A good use case of when humans are much better than AI💋
Virtual forest bathing
Stuck in a concrete jungle? A VR trip to a virtual forest could be just as relaxing as being in the pines.
Beam me up, Scotty 🖖
They say a lot of tech innovations exist because developers got the idea from sci-fi shows like Star Trek. Here's your official captain's log of how the tech in the original 1960s series is progressing, with many theoretical concepts now in the wild.
This one's super Black Mirror. I enjoyed this melancholic AI-generated film about cloning a jazz singer. It's still the uncanny valley, but the music is surprisingly good.
Robots on the ropes
The first rule about humanoid robot fight club 🥊
This little cowboy robot, Rizzbot, is running around Austin, Texas, dishing out compliments. Channelling serious Troy Hawke vibes.
💼 AI business use cases
The AI fixers
Personalised AI messages may give your readers the ick when it feels too smooth or unnatural to be believable. And it's the human consideration and imperfections that make communication worthwhile: "This is me, speaking to you." There are plenty of ways to make AI copy less icky. The irony: Experienced writers and editors know how to do this best.
While AI culls some jobs, it's creating new opportunities for those who know their onions. In a strange circular economy, there's a new role of professional 'AI fixers' – writers and coders cleaning up the chaos and blandification of GenAI projects without any expert oversight, sometimes generating costly mistakes 😲. Where can you lean into your specialism to deliver a better fix?
When it's OK to use AI and for what
The UK civil service has some pragmatic and useful tips on how and when it's OK to use AI in your job application. Identifying your skills = good. Hallucinating new ones = bad. Mid to large-sized organisation: You need to be doing this for all use cases where people are already using free or enterprise AI tools.
Seeking refuge
This one parts the waves: The United Nations is creating AI avatars so people can interact with virtual asylum seekers and learn their stories. Controversial, as they could invest resources into giving voice to real refugees. But could this be a useful tool to experiment with new technologies to build empathy and understanding with new audiences?
AI is reshaping content marketing
The shift to AI as a search tool is shaking up the dyed-in-the-wool methods of content marketing in a good way.
'How to' and generic content are out, as LLMs already know this. It's now about getting more specific about your product information and getting your original and expert research out there. Get rid of those sad gated PDFs on your corporate sites: get your content distributed far and wide. I'm quoted in this informative piece with lots of tasty tips on AI and content marketing.
But contrary to what some may think, web content is here to stay: keeping it accurate is more important than ever. Robert Rose thinks that outdated content on your site could resurface in unplanned and unfortunate ways in AI. Time for a content audit spring clean?🧹
Leaving a legacy
Legacy code like COBOL is a pain in the arse for developers to build on, with few people knowing the right languages to programme it. Morgan Stanley hit on a neat innovation: an AI tool that converts it into a modern coding language. For legacy system coders, like banking, this could be revelatory.
➡️ What to do next to get ahead
1. Invest in AI fluency
Survey your team about their AI training access and usage. Just 20 hours of training could deliver a big productivity boost and move everyone towards AI fluency. Build an inclusive AI adoption strategy so AI benefits and grows the whole organisation.
High AI literacy users are more likely to reject AI than embrace it. Train your people to understand AI's limitations and capabilities. The goal isn't enthusiasm but competence. Smart scepticism beats naive adoption.
2. Be an original
Generic 'how-to' content is dead – AI already knows about it. Focus on creating specific product information, original research and expert insights. Audit your existing content for outdated information. Your content strategy needs to feed AI search, not fight it.
3. Write your AI rules
Before your teams start using AI tools willy-nilly, establish clear guidelines about when and how to use AI at work. The UK Civil Service's approach is a good template. Create policies for all use cases already happening in your organisation.
🔁 ICYMI in Rethinking the Hype Cycle
How to think with AI without losing your mind
Can you still sharpen your brain with the temptation of AI?
Inclusive AI is survival
It's not a revolution if AI leaves half the workforce behind.
AI is the new oil – Rethinking The Hype Cycle #11
AI model access is the new digital divide
Back next Thurs or Fri with a spicy opinion (or thoughtful analysis), and in two weeks with a trends round-up. Sign up to get it first.
Until then, keep it curious 🤔
Susi O'Neill
EVA trust in tech www.evadigitaltrust.com
Tech communications without the hype🎙️Tech talks and inspiration This season’s keynotes 🤙Need this? Get in touch
Such an informative and insightful read, Susi. Thanks!
Glad you found it useful, thanks!