🚀 PR Futurist #102 💊 Impact of AI on consultants 🇨🇳 DeepSeek 🤯 AI Sir Humphrey 🏔️ AI sustainavility 3 February 2025

Welcome

Will you take the red pill or the blue pill?

In this issue, how can we avoid mentioning DeepSeek, the new Chinese AI that has exploded on to the scene. Although at the pace things are changing, it might have already taken over the world while I'm writing this sentence.

We also have a look at the impact of AI on management and PR consulting - will you take the blue pill or the red pill? The UK government has taken the red pill and is implementing 'Humphrey' AI assistants for civil servants. There's also proof that the UK hasn't been totally usurped by American and Chinese AI as video AI company Synthesia raises $180 million at an astonishing $2.1 billion valuation.

If you're concerned about the sustainability of all this AI then Microsoft's Satya Nadella has something interesting to say.

If you want to see the difference between normal AI and AI reasoning, you can try it with both the free and paid consumer versions of Microsoft Copilot without needing to pay OpenAI.

And it's that time of the year again. PRovoke Media's comprehensive annual crisis review is out. This time it has a whopping 28 case studies, including PwC, Google, OpenAI and inevitably Crowdstrike. My personal favourite is Volkswagen. What's yours?

News

Chinese AI DeepSeek upends global AI

You'd have to have been sleeping under a rock not to have noticed the launch of DeepSeek, a new AI from China that has turned AI on its head. The TL;DR is it was built at a fraction of the cost ($5-6 million instead of billions), runs on older lower powered chips and performs as well as OpenAI's top-end reasoning model. And it's open source, which means anyone can download it and run it on their own servers. Lots of people and companies already are. If that's too much, then it has an API (you can connect your tech to it) that's about a tenth of the cost of OpenAI's.

What makes DeepSeek so special is it is capable of reasoning. This means it doesn't just spit out an instant answer but takes time - seconds or even minutes - to ‘think’ about its answer to a hard question in a step-by-step or ‘chain of thought’ way.

It's not all roses in the Chinese AI garden. DeepSeek's privacy policy doesn't provide any privacy or safety for your data which will be saved on servers in China. And its answers are censored by the Chinese government (when it launched I wrote about asking DeepSeek about Tiananmen Square and the Uyghurs). The final flaw is it now appears it was trained on.... wait for it.... OpenAI's own large language modal.

But none of those are insurmountable obstacles. First, as it's open source anyone can run it on a server in a different country with a safer privacy policy.

I'm not convinced there is much difference between planning my holiday with a free version of DeepSeek, where a Chinese company (possibly the government) might access my plans, and using a free version of ChatGPT, where an American company (possibly the government) could do the same. There's nothing useful to either or risky to me.

Second, censorship isn't the sole preserve of China. The US doctrine of 'free speech' so loved by the West Coast tech bros only goes so far. American LLMs are subject to censorship as well. Try asking ChatGPT or Claude about some NSFW (not safe for work) issues, and you might get a curt reply.

Thirdly, it's a bit much for Sam Altman to get upset about DeepSeek using the data in its LLM to create something useful to the world. He's gobbled up most of the copyright data on the internet as the source to create the LLM in the first place.

What's source for the goose is source for the gander - see what I did there?

Our advice is to have a play with DeepSeek. Just do it using personal or public information rather than private or confidential work information.

The prompt for the AI image was the text of this article. It has used the rose garden and the geese!

You can read more about DeepSeek on the BBC and The Guardian or get in touch with us,

Enter the Davos Communications Awards to collect your prize in Davos

Entries to the Davos Communications Awards are open. The winners will be announced at a gala dinner during the Davos World Communication Summit in April. I was the chair of the jury for the first year when the awards relaunched. The entries were impressive and since then have continued to improve.

The Davos Communications Awards "recognise innovation and excellence in exceptional achievements of public relations and communications professionals worldwide, who truly push the boundaries on performance."

Even if you don't enter the awards, you should still book your pass for the summit. Find out more here.

Stuart Bruce

Impact of AI on public relations for PRCA Ireland

Last week Stuart presented a private invite-only webinar for members of the PRCA in Ireland. He talked about the latest developments on the impact of AI on public relations.

AI

How AI will impact management consulting... or PR consultancies?

Are you worried AI might take your job? When ChatGPT first launched it was a worry of many. But a needless one as for most people it couldn't. It could, and will, take many of the jobs of people who don't use it to do their job better.

Reasoning or ‘chain of thought’ ups the stakes significantly. AI is now capable of tackling far more complex questions. Just the sort of thing that companies hire expert consultants to do. Management consultants or public relations consultants.

This article examines the potential impact of AI on consultants.

Be afraid. Very afraid.

Be excited. Very excited.

Will you take the blue pill or the red pill?

Image created with Plugger.ai.

Stuart Bruce

Deep thinking for free in Copilot, no need to pay OpenAI

The exciting news is you don't need to pay OpenAI a monthly subscription or risk your data with DeepSeek to get 'deep reasoning' as Microsoft is giving it away for free. Microsoft has announced is has added GPT-o1 to Copilot for all users, even on the free tier. It now sports a 'Think Deeper' toggle.

Like most things with AI names and licences, it takes some understanding and different articles are explaining it in different ways. Distilling them, it appears:

OpenAI charges $200 per month for access to ChatGPT-o1 in its ChatGPT Pro subscription.
Open AI charges $20 per month for access to ChatGPT-o1 mini in its ChatGPT Plus subscription.
Microsoft offers 'limited' use of the full GPT-o1 in both the free and paid for Plus versions of Copilot.
Microsoft increased usage limits in response to complaints.
Think Deeper still isn't available in the paid versions of Microsoft 365 Copilot.

Stuart Bruce

'Humphrey' AI assistant for UK civil servants

If you're of a certain age, Humphrey will need no introduction. This Humphrey is a set of AI apps to assist civil servants in the UK government.

Parlex is an AI app to predict how MPs will respond to potential policies or parliamentary bills based on what they've said in Parliament and committees. The idea is it will help ministers and civil servants to hone their arguments and messages.

Consult is a tool to read and summarise responses to consultations in minutes or hours, rather than the weeks or months it currently takes. It potentially saves £80 million a year by summarising themes of public consultations automatically rather than relying on dozens of junior civil servants.

Consult is potentially an example of how AI can do the job better than humans. Combing through thousands of responses to identify trends and themes is difficult. Even harder is making sure a golden nugget of a good idea or specific important point isn't missed. AI can potentially do that far better than a team of humans.

Other tools include Minute (a secure transcription service) and Redbox to help prepare briefings and documents.

I'm not entirely convinced by all of these. Parlex is at best a partial solution. If it's only using data from Hansard about what MPs have said in the House of Commons then it's missing far too much vital data. What about their speeches in other places, comments in mainstream media, or on personal social media and websites? Or their contributions to think tanks and other places they might talk about policy?

This would provide real insight and the AI predictions of their potential responses would be worth thinking about.

Actually that level of insight into an MP's thinking is possible right now without the need for Parlex.

If you're wondering who Humphrey is then he is a character from the popular BBC sitcoms Yes Minister and Yes Prime Minister, where he serves as a senior civil servant. It's rather an odd choice of name for the AI tools as Sir Humphrey Appleby is best known for being a master of obfuscation which he used to manipulate his minister and later prime minister. He was famed for in every episode making long-winded statements designed (usually successfully) to confuse and fatigue the listener.

Hopefully the AI Humphrey won't obfuscate as much as this Sir Humphry confounding his minister.

I'd have called the AI tools Dorothy, but you'll have to be even more of a Yes Prime Minister nerd to understand why.

Stuart Bruce

Microsoft's Satya Nadella on AI sustainability

One of the most frequent criticism levelled at AI is sustainability. AI guzzles electricity and water (for cooling). Although that might be one of the things that DeepSeek has upended as if it can really run on older low power chips then it doesn't guzzle energy and water after all.

Sustainability is a justifiable criticism, but it's also a solvable problem which technology companies are already well on the way to solving.

In this video, filmed at the World Economic Forum in Davos, the CEO of The Atlantic Nick Thompson chats to Microsoft CEO Satya Nadella about a proposed new metric for measuring the efficiency of AI models: tokens per watt per dollar. Thompson says it’s a smart framework for how to bring energy costs into the way we track intelligence.

Stuart Bruce

Improving news engagement with AI-powered personalisation

Publishers are starting to embrace AI to improve engagement with readers. PR professionals need to understand this as it impacts on media relations. They also need to understand how they can use similar ideas for owned content.

Stuart Bruce

UK AI video company Synthesia raises $180 million at $2.1 billion valuation

We don't hear enough about UK AI firm Synthesia. It enables you to create videos from text using digital humans. You can choose from lots of different realistic video avatars. Its more expensive plans let you clone yourself and create a personal video avatar.

The company has just raised $180 million, raising its valuation to $2.1 billion, making it one of the largest AI start-ups in the UK. Business Insider has published the pitch deck from Synthesia's recent successful funding round.

I created a video avatar of myself using Heygen, a US alternative to Synthesia that offers more affordable pricing for personal video clones.

Stuart Bruce

Get in touch

Are you at risk because you don't have an AI policy?

Whether you have a policy or not, people in your organisation will be using AI. Our Global CommTech Report with PRovoke Media revealed that only 40% had a policy, while 66% used personal AI tools at work.. The release of DeepSeek increases the risk as your data is stored on servers in China and its privacy policy provides no protection.

At Purposeful Relations, we can help you guard your organisation's data and reputation by helping you to develop an AI policy that encourages innovation and experimentation - safely. For advice, please contact us.

Crisis communication

PRovoke Media's annual crisis review: PwC, Google, OpenAI, Ticketmaster and of course Crowdstrike

I'm a big fan of PRovoke Media's annual crisis view which summarises the major crises of the previous year with expert insight and analysis. It's usually in three parts, but this year has a whopping 28 case studies in four parts.

The crises this year are Harrods, Planet Fitness, Wisetech, Baidu, SingPost, G4 Educação, Samherji, Jaguar, TotalEnergies, Petrobras, PwC, Unilever and Ben & Jerry's, Chiquita, Zelle, Google, Heinz, OpenAI, Volkswagen, Qantas, AstraZeneca, Starbucks, United Healthcare, TikTok, Boeing, Ticketmaster, McDonald's and Crowdstrike.

My favourite is probably Volkswagen as it's a great reminder that in a crisis it is usually impossible to satisfy every stakeholder. It's critical to have a clear objective and know which stakeholder or stakeholders are the most important, even if it means damaging relationships and reptuations with others.

I incorporate these case studies into crisis communications training for my clients at Stuart Bruce Associates and in third-party courses for the CIPR.

Stuart Bruce

Professional practice

CIPR unveils new 2025-29 strategy

The Chartered Institute of Public Relations has revealed its new strategy for 2025-29. It doesn't contain anything radically new and is a continuation of the great work it has already been doing.

It focuses on four key priorities: Chartership and Lifelong Learning; Leading Practice Development; Advocacy; and fostering a Diverse and Collaborative Community.

One of the key initiatives planned for 2025 is to continue raising the profile of chartered status. As a chartered member this is one I particularly welcome. It's most important to do this with senior leaders who will be employing public relations professionals or hiring consultancies.

We need to see recruitment ads, job specifications and RFPs (request for proposals) listing PR qualifications, chartership and proof of CPD (continuous professional development) as standard requirements.

Stuart Bruce

Digital and social

NFL teams banned from Bluesky

What can be more American than the NFL and free speech? Although apparently the two aren't compatible. Far from embracing free speech the NFL (National Football League) is clamping down on the wrong type of free speech.

The NFL has told the New England Patriots to shut down the team's Bluesky account. The NFL apparently wants to keep its cosy relationships with Elon Musk's X and Mark Zuckerberg's Meta (Facebook, Instagram and Threads) rather than embracing a more open platform.

Stuart Bruce