Chris Tubb
@christubb.com
Intranets, digital workplace, internal comms, search, findability, strategy, governance, user journeys, user and stakeholder research. AI heretic sparktrajectory.com Consulting unit. Brighton.
Pinned
Chris Tubb
@christubb.com
· Dec 2
Hello, I'm Chris. All sorts of intranet, digital workplace and internal comms useful since 1996 and now partner at Spark Trajectory in UK. 17 years on the other place but it became a read-only medium when everything all got a bit fraught. Here to share and find interesting people and ideas.
What’s so interesting about this grotesque and macabre idea (do check the quoteskeets) is that there must be a complete absence of market research in the tech world.
Meta has patented AI that can run a dead person's account, continuing to post and chat on their behalf
It can message and video call by replicating a user's online behavior using their past data
It can message and video call by replicating a user's online behavior using their past data
February 17, 2026 at 1:35 PM
What’s so interesting about this grotesque and macabre idea (do check the quoteskeets) is that there must be a complete absence of market research in the tech world.
IBM corporate training documentation from 1979
February 12, 2026 at 4:54 PM
IBM corporate training documentation from 1979
I am worried that the folks I see on LinkedIn doing this are doing it for free.
Microsoft, Google, Meta and Amazon paying influencers $400k-$600k to post months of AI hype content. Absolutely busted, failed industry, embarrassing era
www.cnbc.com/2026/02/06/g...
www.cnbc.com/2026/02/06/g...
Google and Microsoft offer lucrative deals to promote AI, but even $500,000 won't sway some creators
Tech companies including Google and Microsoft are paying influencers hundreds of thousands of dollars to promote AI products.
www.cnbc.com
February 11, 2026 at 7:52 PM
I am worried that the folks I see on LinkedIn doing this are doing it for free.
To the UX people on the purgatory site blaming AI adoption in organisations on some form of cultural resistance and not the shortcomings of the tools themselves. Shame on you.
February 10, 2026 at 6:26 PM
To the UX people on the purgatory site blaming AI adoption in organisations on some form of cultural resistance and not the shortcomings of the tools themselves. Shame on you.
I don't wanna be all media-is-biased-and-all-that but today Reform has said they would ban working from home and defund a university because of its debating society and the news is all "Is Starmer XYZ?"
February 10, 2026 at 4:23 PM
I don't wanna be all media-is-biased-and-all-that but today Reform has said they would ban working from home and defund a university because of its debating society and the news is all "Is Starmer XYZ?"
For completeness on me socials (and @thompsonsimon.com @sharonodea.com and @lisariemers.bsky.social who actually hang) here's the last post of my series about what we are actually going to do about AI in the enterprise after the hype has died down. www.linkedin.com/pulse/ai-epi...
The AI Epiphanies, Part Three: Mapping and managing the risky epistemic landscape
We argue organisations must map employee tasks and journeys to manage epistemic risk and we have come up with a full solution to do so.
www.linkedin.com
February 10, 2026 at 9:21 AM
For completeness on me socials (and @thompsonsimon.com @sharonodea.com and @lisariemers.bsky.social who actually hang) here's the last post of my series about what we are actually going to do about AI in the enterprise after the hype has died down. www.linkedin.com/pulse/ai-epi...
Read enough AI-generated LinkedIn posts and the rhythm becomes a modern shave-and-a-haircut-two-bits. It's hilarious.
February 4, 2026 at 4:40 PM
Read enough AI-generated LinkedIn posts and the rhythm becomes a modern shave-and-a-haircut-two-bits. It's hilarious.
What the world really needs is a LinkedIn post about "Learning to say no to yourself about posting about learning to say no."
February 3, 2026 at 2:32 PM
What the world really needs is a LinkedIn post about "Learning to say no to yourself about posting about learning to say no."
Post in which I propose a new layer of information management to deal with the sprawl of enterprise AI and the inevitable misinformation. www.linkedin.com/pulse/ai-epi...
The AI Epiphanies, Part Two: How Provenance restores order to a fractured digital workplace
In my previous post in this series I looked at some of the fundamental constraints on making enterprise information AI legible in ways that will let us do some of our work for us and be able to retrie...
www.linkedin.com
February 3, 2026 at 9:05 AM
Post in which I propose a new layer of information management to deal with the sprawl of enterprise AI and the inevitable misinformation. www.linkedin.com/pulse/ai-epi...
AI in the enterprise isn't going to turn out like you think. You need the field of Provenance before you get called the Chief Misinformation Officer. www.linkedin.com/pulse/ai-epi...
The AI Epiphanies, Part One: Seeing the Hidden Landscape Beneath Enterprise AI
The future of work isn't going to turn out like you think. Not only are the technologies of Large Language Models grossly misunderstood by vendors and organisations alike, the very nature of informati...
www.linkedin.com
January 27, 2026 at 8:54 AM
AI in the enterprise isn't going to turn out like you think. You need the field of Provenance before you get called the Chief Misinformation Officer. www.linkedin.com/pulse/ai-epi...
So this is normal now.
Looks like LLMs are *very* vulnerable to attack via poetic allusion: "curated poetic prompts yielded high attack-success rates (ASR), with some providers exceeding 90% ..."
https://arxiv.org/html/2511.15304v1
https://arxiv.org/html/2511.15304v1
November 20, 2025 at 9:02 PM
So this is normal now.
One of these is not like the others, Apple Music. Bangers for sure, but in 1827.
November 19, 2025 at 3:20 PM
One of these is not like the others, Apple Music. Bangers for sure, but in 1827.
Bluesky idea: Restore human-only signals. When you post, your phone scans your face. Result: botdeath.
November 14, 2025 at 3:33 PM
Bluesky idea: Restore human-only signals. When you post, your phone scans your face. Result: botdeath.
This is why AI isn't going to turn out like businesses think. Like a toddler they believe that the environment won't respond back.
When an AI-powered writing tool was rolled out on a job site, the length of proposals exploded. Signals employers used to identify good candidates — like quality of writing, and relevance of experience — became ubiquitous. That was OK for bad candidates, but terrible for good ones.
November 14, 2025 at 12:08 PM
This is why AI isn't going to turn out like businesses think. Like a toddler they believe that the environment won't respond back.
Claiming AI agents have autonomy is like me claiming my oven roasts a chicken on its own, between me putting it in and taking it out.
October 24, 2025 at 10:07 AM
Claiming AI agents have autonomy is like me claiming my oven roasts a chicken on its own, between me putting it in and taking it out.
Duolingo’s increasing desperation to sell me their AI enhanced version is getting kind of off putting. Mates, make it part of the core offering or move on.
October 23, 2025 at 6:44 PM
Duolingo’s increasing desperation to sell me their AI enhanced version is getting kind of off putting. Mates, make it part of the core offering or move on.
It's fascinating to me about how the intentions of the AI companies (fire all the people) and the uptake of AI in business (get it to do the work of people we fired already) are in conflict.
October 16, 2025 at 9:58 AM
It's fascinating to me about how the intentions of the AI companies (fire all the people) and the uptake of AI in business (get it to do the work of people we fired already) are in conflict.
New poster boy just dropped.
October 16, 2025 at 8:43 AM
New poster boy just dropped.
When the AI crash comes, there is going to be an awful lot of "Poor me my smol bean startup has died we were only trying to replace thousands of actual jobs."
October 15, 2025 at 8:48 AM
When the AI crash comes, there is going to be an awful lot of "Poor me my smol bean startup has died we were only trying to replace thousands of actual jobs."
My final AI heresies post over on the purgatorysite, in which I politely ask large organisations whether shoving AI into all their projects will actually always work (no) and whether it's a bubble ready to pop (yes).
www.linkedin.com/pulse/ai-her...
www.linkedin.com/pulse/ai-her...
The AI Heresies, Part Six: Strategy and Choosing the Right Problems in the Digital Workplace
Throughout the previous five posts, (part one, part two, part three, part four, part five) I hope that I have aired my heretical inclinations when it comes to our current version of what we call AI. I...
www.linkedin.com
October 14, 2025 at 8:22 AM
My final AI heresies post over on the purgatorysite, in which I politely ask large organisations whether shoving AI into all their projects will actually always work (no) and whether it's a bubble ready to pop (yes).
www.linkedin.com/pulse/ai-her...
www.linkedin.com/pulse/ai-her...
Reposted by Chris Tubb
“Current architectures of LLMs cannot imagine, but they can sequence … For the same reason that a dog can go to church but a dog cannot be Catholic, an LLM can have a conversation but cannot participate in the conversation.”
I've been re-acquainting myself with NNs, transformers and LLMs this week, so these are early thoughts. But I am starting to think about language without the capacity to imagine language, i.e., an LLM unable to imagine itself participating in language. mail.cyberneticforests.com/what-machine...
What Machines Don't Know
Imagining Language Without Imagination
It's important to acknowledge that Large Language Models are complex. There's an oversimplified binary in online chatter between the dismissive characterizatio...
mail.cyberneticforests.com
October 12, 2025 at 8:21 PM
“Current architectures of LLMs cannot imagine, but they can sequence … For the same reason that a dog can go to church but a dog cannot be Catholic, an LLM can have a conversation but cannot participate in the conversation.”
Won’t be long now…
‘The hundreds of billions of dollars companies are investing in AI now account for an astonishing 40 per cent share of US GDP growth this year… Outside of the AI plays, even European stock markets have been outperforming the US this decade’
on.ft.com/4pTQ3US
on.ft.com/4pTQ3US
America is now one big bet on AI
It’s seen as the magic fix for every threat to the US economy
on.ft.com
October 6, 2025 at 7:26 AM
Won’t be long now…
Part five of my heretical take on digital workplace AI over on the purgatorysite, where I discuss what AI needs in the way of governance and why we are going to need a lot more and not a lot less. www.linkedin.com/pulse/ai-her...
The AI Heresies, Part Five: Governance Eats the Gains
Governance is the dull magic you need to manage anything. Governance = (Clarity + Meaning + Accountability) x Attention.
www.linkedin.com
October 2, 2025 at 8:03 AM
Part five of my heretical take on digital workplace AI over on the purgatorysite, where I discuss what AI needs in the way of governance and why we are going to need a lot more and not a lot less. www.linkedin.com/pulse/ai-her...
Nice illustration of AI and Jevons' paradox. AI makes the amount of cases in Brazilian legal system go up.
“We note that the use of AI, in the end, rather than diminishing litigation, is increasing it…[AI] may be a solution, but no one’s sure if it will actually work.”
AI is helping judges to quickly close cases, and lawyers to quickly open them
Brazil’s overburdened courts and lawyers are adopting artificial intelligence. But experts wonder whether it serves justice.
restofworld.org
September 29, 2025 at 10:05 AM
Nice illustration of AI and Jevons' paradox. AI makes the amount of cases in Brazilian legal system go up.
