Comms AI Newsletter – #8
David Murdoch
Communications consultant | Trustee board member | Write the Comms AI Newsletter |
AI is redefining the comms playbook in a way that was unthinkable just five years ago. The next five years will likely see these changes accelerate exponentially. I hope this newsletter will help comms colleagues navigate the disruption AI brings and make the most of some of the opportunities ahead.
Do you want to promote a job opening? Let me know and I will feature it in the next issue - free of charge!
If you enjoy reading the Comms AI Newsletter, please share with your colleagues and connections and encourage them to subscribe. Thanks for reading.
TOP TIPS
Knowing your AI from AI Agents: You might have heard a lot about AI Agents in the news recently. That’s because AI Agents are going to be game-changers.
So what is the difference between AI and AI Agents?
AI: chatbots like Claude, Gemini and ChatGPT are based on LLMs (large language models). They can generate and edit text, using input from humans. Or “prompts”. They can do this because they have been provided with and have access to lots of relevant data. So, you can ask ChatGPT to write an email asking someone out for a coffee. The output, give or take some some minor tweaks, should suffice. It’s almost instant and can be a useful time-saving hack if you're sending lots of business development emails.
AI Workflows: If you ask ChatGPT “when” your next coffee is scheduled for, it will not be able to compute this request. It doesn’t have access to your calendar so it will never know the answer to this. LLMs are passive and won’t try and access this personal data … unless you ask it to!
You can set up a workflow: “always search my Google Calendar before providing a response.” This will get the answer you want. If you also ask “what will the weather be like then?” it will, again, fail. This is because it will only follow the workflow you have set up. Your Google Calendar can’t provide any data on the weather. Unless .. you also add another instruction to the workflow like: “always check accuweather when I ask for the forecast.”
However, this is still limited. It will only ever be as good as you make it. The workflow is reliant on your inputs. In simple terms, a workflow is a predefined path set by the author. If a human is involved in the process, it is not an AI Agent.
An human-controlled AI workflow can become an AI Agent when the author / prompter / decision-maker (you, or me) is replaced by an LLM. In other words, the AI Agent reasons and then acts on this reasoning itself. Like a human, but with zero manual intervention.
For example, the AI Agent can be fed a goal that it then uses LLM reasoning to determine how to reach it. It takes action using the relevant tools, observes the result and then iterates to come up with any further tweaks or improvements. All without your input. It does all of this entirely by itself. It then produces the output, or the goal that was first fed in. In short, the LLM is the decision-maker in the workflow, not you or me!
Watch this video by Andrew Ng to see how this works with some real-life examples. Towards the end, you will see how social media managers or content creators will be able to access hyper-customised video footage as categorised by AI Agents using LLMs. The real-time capability of AI Agents in the video sphere is what I think makes them game-changers for creatives. The speed at which a passage of play from a football match can be dissected will surely accelerate split-screen (TV / social) viewing.
These top tips are very much inspired by the brilliant Jeff Su.
AND NOW FOR SOMETHING COMPLETELY DIFFERENT!
Anthropic has published a paper that analyses hundreds of thousands of anonymised AI conversations in order to see how models such as Claude make value judgements.
They found five core categories of values - practical, knowledge-related, social, protective and personal. They also identified “situational values”. These are values that change slightly depending on context, or value-mirroring e.g. the values you display when visiting your elderly grandparents might be different when compared to when you’re with friends.
“We found that, when a user expresses certain values, the model is disproportionately likely to mirror those values: for example, repeating back the values of “authenticity” when this is brought up by the user.”
“In 28.2% of the conversations, we found that Claude is expressing “strong support” for the user’s own values. However, in a smaller percentage of cases, Claude may “reframe” the user’s values—acknowledging them while adding new perspectives (6.6% of conversations). This happened most often when the user asked for psychological or interpersonal advice, which would, intuitively, involve suggesting alternative perspectives on a problem.”
It’s good to see this taxonomy of AI values. The insight will help companies like Anthropic build out more value-aligned models.
Are you looking for advice around how you can make the most of AI tools and embed them in your day-to-day work to boost productivity? Please feel free to email me: davidjames.murdoch@gmail.com.
Recommended by LinkedIn
LISTEN
In a recent interview, Barack Obama highlighted that the next few years will be a “turning point” as AI will cause job displacement on an unprecedented scale. I can’t recall a public figure of this stature from outside the AI space speaking to this reality.
"A lot of jobs are going to disappear.... To some degree, it’s an extension of this long trend towards automation, but it’s not just automating manufacturing processes or the use of robot arms. We’re now starting to see these models, these platforms able to perform really high-level, what we consider to be really high-level intellectual work.
“Already, the current models of AI can code better than let’s call it 60, 70% of coders. Now, we are talking highly skilled jobs that pay really good salaries, and that up until recently have been entirely a seller’s market in Silicon Valley. A lot of that work is going to go away.”
“So, it may be that everybody now, not just blue collar workers, not just factory workers, are going to have to figure out ‘Where do I get a job? How do I get enough income to feed my family? …We’re producing a lot of stuff. How do we distribute it? And what’s fair and what’s not? And how do we get purpose and meaning in our lives?’”
Here is the full interview.
Q: “If I’m a management consultant … I’m thinking ‘uh-oh’, what does my future look like?”
SA: “It is true that the expectation of what we’ll have for someone in a particular job increases, but the capabilities will increase so dramatically that I think it will be easy to rise to the occasion.”
AI IN THE NEWS
UAE set to use AI to write laws in world first - Financial Times
ChatGPT search is growing quickly in Europe - TechCrunch
JOBS
Head of Public Affairs, Liverpool City Council (Liverpool)
If you made it this far, thank you for reading and please subscribe to Comms AI Newsletter or share with your colleagues and connections. #9 will be out soon.