Bob Pickard

Principal @commsleadership • Public relations for newsmakers • PR career @ 🇺🇸🇨🇳🇯🇵🇨🇦🇰🇷🇸🇬 Battle-hardened communicator • On the record • Asia hand 🌏

[Bob Pickard @BobPickard]

Algorithmically synthesized reputation:
There is no such thing as a single public-domain version of reputation.

For a long time, reputation seemed a more straightforward matter of checking Google page one results or doing a perception audit of stakeholders. There's never been a single 'official reputation' for individuals or organizations. But today's AI-era reputation is becoming much more fluid and multiplexed compared to the past.

Ask ChatGPT, Claude and Gemini about the same person or organization and you can get materially different answers. Each LLM draws on different sources, weighs them differently and resolves ambiguity in its own way. The same system will vary again according to the model version, the wording of the question, and the day it is asked.

For public relations, this changes the management of perception in an important way. We have long distinguished between image programmed, what a name deliberately communicates about itself, and reputation earned, what others conclude from its conduct over time. Generative AI now takes both, combines them with journalism, public records, criticism and historical material, and produces what might be called an algorithmically synthesized reputation.

That reputation is not sitting somewhere like a media monitoring database waiting to be retrieved. It is being rebuilt on demand, and different systems assemble it from different portions of the same information environment.

Muck Rack's May 2026 study of more than 25 million links cited by ChatGPT, Claude and Gemini across 17 industries found that what it classifies as earned media drives 84 per cent of citations, with journalism alone at 27 per cent and paid or advertorial content at 0.3 per cent. The same research shows how differently each system behaves. ChatGPT cites sources in 96 per cent of its responses, but averages five citations. Claude cites in only 55 per cent, and averages thirteen when it does.

The single most-cited domain is Wikipedia on ChatGPT, PubMed Central on Claude and Reddit on Gemini. Smaller samples point in the same direction: BrandGhost's AI Discovery Observatory reported only a 16 per cent source overlap in its August run among the LLMs it could compare directly (with half of comparable recommendation sets sharing no brands at all).

One 'favourable' ChatGPT answer therefore tells us only a small part of the story. What matters is the range of representations being generated across systems, which sources are producing those conclusions, where outdated or inaccurate material remains influential, and whether the public evidence surrounding a person or organization is strong enough to support an accurate account.

This makes genuinely earned reputation more consequential. Owned content can be produced in almost unlimited quantities. Credible independent evidence, authoritative third-party validation and a consistent record of actual behaviour are scarce by comparison, and that scarcity is precisely what gives them weight when a machine goes looking.

The practical question is what independently verifiable material exists about you, how current it is, and whether it holds up when a system reconstructs your reputation on someone else's behalf.

[Bob Pickard @BobPickard]

@PeteBritish This is an excellent video clip BTW – many thanks for sharing.

[Carole MacNeil @carolemacneil]

I worked at CBC as a journalist for decades. We've had the conversation about the word "terrorism" a few times. It was never about Islam, colonizers or anything like that. Most of the time, it came down to the memory of Nelson Mandela and the ANC. Mandela and the African National Congress were designated terrorists until 2008. How should media have dealt with that? This is the example that was often cited and likely contributed to CBC policy of many years to leave it to attribution.

[Bob Pickard @BobPickard]

In retrospect, I was spectacularly fortunate to have had that experience, and I am grateful for the memory. I remember passengers smoking on the flight (!), the champagne and caviar cart, and a visit to the cockpit, which was filled with analogue-era mechanical switches and gauges. I also recall looking out the window at a dark-blue, almost purple sky, seeing the curvature of the Earth, and putting my hand against the window and feeling the warmth created by aerodynamic heating at supersonic speed.

[Bob Pickard @BobPickard]

@TheRickWilson @The_Johnleb @sundarpichai This is how it looks in Canada this morning. Google puts the asinine (Lake America) in brackets, and on the American version it doesn't look like Google even does that for the longstanding (Lake Ontario) recognized by everyone else worldwide except for the Trump regime.

[Laura Babcock 🇨🇦 @LauraBabcock]

It was a great one! https://t.co/znrPdsyTb3

[Bob Pickard @BobPickard]

@android21539794 @sarobertson_ And you are machine account @android21539794, another profile with a long alphanumeric name and no real person's name or photo.

[Bob Pickard @BobPickard]

@Royal23Centrism Awareness is key, eh?

[Bob Pickard @BobPickard]

Social media algorithms are no longer simply deciding what information we see. They may also be helping shape the emotional state in which we receive and interpret communication.

A new University of Texas at Dallas study found that people with more depressive symptoms were served more depressive content by their personalized TikTok and Instagram feeds, with corresponding differences in emotional processing. The researchers describe a potentially self-reinforcing loop: emotional state influences engagement, the algorithm learns from that behaviour, and the resulting feed may reinforce the state that helped produce it.

For public relations, that matters because audiences do not encounter messages in a psychologically neutral environment. Algorithms may already be influencing mood, directing attention and shaping receptivity before a message from an organization or public figure even appears, affecting not only what people see, but the state of mind in which they interpret it.

[Bob Pickard @BobPickard]

@RememberingOct7 @sarobertson_ Another "you guys" post. It's almost like multiple accounts are scripted and coordinated by a foreign actor to sow division and spread poison in the Canadian social media sphere, for some reason.

[Bob Pickard @BobPickard]

@GareyCozad @sarobertson_ You don't follow the news, American friend? These insane tariffs are a massive economic attack.

[Bob Pickard @BobPickard]

@HeartCanadaIRL @sarobertson_ You guys?

[Bob Pickard @BobPickard]

@kilmaltie @sarobertson_ 🇺🇸

[Bob Pickard @BobPickard]

Psychologists examine downside of social media algorithms
https://t.co/0qhUiyT0sR

[Bob Pickard @BobPickard]

For some reason, I notice that your account has never posted before, but today you are busy making comments on the posts of several people calling them "retards" or "fkn retard" and so on. This seems to be the sole purpose of your account. Have you considered using your time on social media more constructively?

[Bob Pickard @BobPickard]

@DustinEBissell @sarobertson_ I was genuinely amused to review your social media profile just now. You are a very interesting man.

[CHCH News @CHCHNews]

As of today, the Saturday Evening News at Six and 11 are cancelled.

CHCH News Director Greg O'Brien (@gregobr) has the details on why the decision was made.

https://t.co/TrYDRt7h2N

[Bob Pickard @BobPickard]

@stevenb63497069 @PrimalReality @sarobertson_ Those are the countries where I've been fortunate to live and work. There are millions of Canadians overseas and collectively they comprise an important element of this country's soft power potential.

[Bob Pickard @BobPickard]

@MarcTotes @sarobertson_ Astute analysis. I hope and suspect that you are right.

[Bob Pickard @BobPickard]

@JDW030138253456 @sarobertson_ Thanks for your thoughtful and well considered machine-generated message of projection and toxicity, JDW030138253457.