The fifth update to our “every question we could think of” guide
We originally wrote this guide in December 2022, and it didn’t age well. AI is evolving rapidly.
Now in the fifth major revision & rewrite we’ve updated every answer and added new questions to address topics such as proven AI benefits, impacts, costs, learning curves and more. We have also expanded the guide to look at new capabilities such as video, infographics and planning.
This information is based on conversations that we have had with over 100 organizations plus the fact that we live and breath the topic of AI every day. You may also notice that many of our answers are less direct that they used to be – this is because the topic is becoming much more nuanced as it matures and as the conversation moves from “what is it” to “how do we use it”. We’ve tried to give our most honest answers.
Should you find that one of our answers hints at something but is not concrete, then it may be because we are trying to avoid selling here. But do reach out and we will be happy to give you all details – we believe that everything here is solvable!
Do you have different or additional questions? Please let us know – we are happy to include them.
This guide contains the following sections:
- The status of AI in communications and marketing today
- Understanding AI and its capabilities
- AI Applications in Communications and Marketing
- Implementing AI – the hard part
- The future and the road ahead
The status of AI in communications and marketing today
Where are we on the AI journey and how is it shaping up? We’ve tried to address some of the more urgent questions upfront.
Q: Now that the hype has settled, are we seeing real business benefits from AI – particularly in Communications and Marketing?
The benefits from AI are very real and very significant. There are many AI tools that can measurably save time, boost creativity and increase impact.
However there is a difference between the existence of benefits and the leverage of benefits. Few (but definitely some) companies have made the commitment and investment required to really leverage AI, so from that perspective the race is wide open.
Q: Can AI already make a significant impact on our communications today?
Yes. A lot has happened since ChatGPT introduced the world to AI in 2022. See above – “There are many tools that can measurably save time, boost creativity and increase impact“. AI solutions have become far more concrete and have evolved beyond simple image and text creation.
Q: How much do we have to think about investing? Is it expensive?
In reality the biggest investment required right now (and the scarcest resource of all) is time. It takes some time investment to understand and think through the opportunities and the road to implementation.
However, we think that looking at cost is to look through the wrong end of the telescope. The business case for AI is significant, particularly in comparison to what is typically spent on today’s production and technology services. That said, it also doesn’t have to be expensive.
Q: There are so many solutions, so many tools and so much change. Is it worth waiting until things settle down?
We are still in the early phases of this technology and much will still change. However, some of the fundamentals will not change, such as the lasting need to rethink the concept of what the humans should do and what the machines can do. At the same time, technology adoption (and hence the risk of being out-paced by competitors) tends to be exponential – so we don’t see much upside from waiting.
Q: We have limited budgets and resources. AI seems out of reach for us?
We appreciate this problem – many people face this. We still think that the business case for time and cost savings through AI is significant, particularly if you face severe resource limitations.
The challenge is that there is little generally accepted wisdom and no standard playbook for the best way to approach this quandary.
Our best advice is to take a very critical look at where time and money go today, and then have a detailed look at how AI could re-write this equation for you.
Q: Some of our team members are afraid / nervous / against using AI. How can we address these concerns?
Research suggests that those who are most against AI are also the least familiar with it. Taking a little time to demonstrate, familiarize and discuss the benefits of AI tools is usually helpful. Exposure to AI can show how it can be applied to real problems and demonstrate how much AI is still dependent on human input.
See also: Will AI take my job?
Q: Our IT team is responsible for AI deployment in our organization. So I don’t have to worry about it?
We are sure your IT is excellent, but here a few things to consider:
AI is not like any other generic productivity tool. AI is a fundamental strategic change that at some point will be core to your process and your team’s competence. We would be very careful about completely delegating such critical changes.
Secondly, although tools such as ChatGPT are generic productivity tools that can be applied to many problems, experience already suggests that AI works best when applied to a specific problem and a specific data set. What we mean by this is that getting real value out of AI requires very specific knowledge and understanding of your need, process and existing skills – you may want to check that your current IT approach supports and understands this.
Q: We are a Microsoft shop and we are deploying Microsoft Copilot. So we are all covered?
Copilot is a great solution, but be aware that it is a generic solution.
What do we mean by this?
For example, Microsoft Excel is a generic productivity solution. Theoretically with a little time and effort it can do anything that Salesforce can do. But people still need and use Salesforce due to its benefits in being a very specific and tailored solution.
In the same way, solving very specific communications & marketing challenges can also require specific applications built to solve specific issues and challenges.
See also “Do we need anything other than Chat GPT?“
Understanding AI and its capabilities
AI has taken big leaps forward in recent times. How good is it really, how seriously do we need to take it, and does it live up to the hype?
Q: How good is AI at creating real content?
Good is a subjective term, so let’s compare against the average which is easier to gague.
AI’s ability to write grammatically correct and coherent text is way above average. AI’s ability to reason and evaluate arguments is also way above average – in most cases. And AI is faster than average.
AI’s ability to be confidently incorrect is also a long way above average while its awareness of its limits is very far below average (although the average may be debated). See also What about situations where AI just makes things up?
What AI mostly lacks is the subtle context that makes up a big part of getting the right message across – although things are rapidly improving in this regard.
Right now this makes AI the perfect tool to tackle time consuming research, planning and briefing tasks, leaving plenty of room for superior human judgement and nuance. We call this approach AI Accelerated Content.
Q: But is AI content “good enough” yet?
It depends what you want to use it for, but in the average case and if properly instructed AI content is good enough. If not for the final draft, certainly for the edit ready draft.
In many cases it is better than good enough. Although in some it is admittedly terrible.
And there are also many cases where AI content is far superior than human content.
The best approach here is to do a little light testing against your specific use cases. In many cases you will probably be positively surprised.
Q: How quickly will AI continue to improve?
We believe that the fundamental AI models will evolve incrementally rather than exponentially, so we expect to see consistent but frequent and continuous improvements over the next year. (but we don’t discount any surprises)
We also think that the models are not the limiting factor today – we see a lot of untapped potential to build even more useful solutions based on existing models.
In short, AI will continue getting better and better for some time.
See also: How will this technology develop and what will its impact be in 1, 2, 5, 10 years?
Q: What about images?
Image generation has improved significantly over the last year and can replace stock or bespoke photography in many cases today. In some cases AI generated images are even better than the real thing since the only limit is on the user’s imagination rather than the availability of imagery.
However, like text generation, image generation still has some practical drawbacks for production quality usage. In some cases the aesthetic is clearly quite “AI”, in complex scenes there can be challenges with the interaction between objects, and there are plenty of cases where you just can’t quite get exactly what you need.
Q: And Videos?
New models such as Sora have dramatically improved the capabilities of AI generated video, although we are still far from replacing today’s video shoots with AI.
While the pace of improvement has been dramatic we have little current insight into how steep the improvement curve will be from here. We wouldn’t be surprised if we are much closer to replacing some video shoots in the next 18 months.
Q: And infographics?
Infographics represent another level of complexity that AI is not quite ready for yet. But we expect that to come sometime soon.
However, what some AI tools can already do is take a complex set of inputs and create an infographic script that a designer can quickly implement – which is useful since the research and scripting are usually the most intensive part of of the process.
Q: Can AI be used in other areas, such as media monitoring or media relations?
Media monitoring solutions have been claiming to be AI based for some time now. Mostly these approaches are superficial enhancements to existing monitoring and do not add significant improvements in speed or quality.
However, new approaches are emerging which take a fundamentally new approach to the problem, such as AI agent based approaches that simulate human media consumption to provide a more accurate representation of perception. Watch this space.
Q: What about planning?
Yes!
Systematic communications planning and optimization requires detailed audience and topic analysis – which is a great application for AI.
What we are saying here is that we know some people who have excellent solutions for both planning and for media monitoring..
Q: I asked ChatGPT a question and it was completely wrong. Can I take AI seriously?
Hallucinations, as they are known, can be a common problem. Large Language Models (LLMs) are designed to be helpful assistants and often try to be too helpful by just filling in any knowledge gaps. The model will take a straight line from question to answer, making up any missing details along the way.
However there are solutions designed to eliminate hallucinations by using base language models as a reference for language construction and by relying on your own data and briefing information to provide actual knowledge and source writing material. This is based on the principle that most organizations already have more than enough basic content – the challenge is to continually reshape it into the form in which it is needed.
Q: Will our audience know when something has been written with AI?
Content written with large language models such as Chat GPT can feel a bit … synthetic. But it is possible to train AI systems on a specific voice and to set clear style guidelines. Providing that AI with rich briefing material also helps to create content that feels fresh, natural and highly original. The secret is in giving the AI enough quality data to work with such that it can create something unique.
Q: Can AI be creative?
The definition of “creative” can be quite subjective. Today’s AI isn’t going to pull completely new thoughts out of the air and create brand new ideas and concepts. But AI excels at a common form of human creativity – combining existing knowledge into new and original thoughts.
In fact, we find that AI’s ability to quickly parse large amounts of information gives it a unique ability to make completely new and original things out of old parts. The proof of this often occurs in looking at a piece of AI content and saying “wow – that’s interesting..”
Our experience shows that this can be particularly valuable in planning processes where AI is able to bring a large amount of data and many different concepts together to form new ideas and approaches.
Where AI typically struggles against its human counterparts is in being able to bring broader perspectives to the creative process. Although AI can work with vast amounts of data it is always limited in comparison to the broad every day human experience consisting of many relationships, observations, experiences and perspectives which means that it can lack context and therefore judgement.
For example, because of this lack of embodiment AI can say “it feels sad” without actually having experienced sadness. We will leave that to a longer philosophical discussion.
Q: AI seems to be changing so fast. How can we keep up?
This is a hard question to answer because AI is changing even faster than the number of consultants, commentators and opinions that fill our feeds every day.
That’s why we have been continually updating this Q&A (and why are are embarrassed by how quickly and badly it dates). – This is already the fifth complete rewrite of this document in 1 year.
Q: How can we measure the performance or success of AI applications in marketing and communications?
AI generated output can be measured in the same way that other content is measured – does it create metrics that show positive business impact?
However, unlike traditional approaches AI can ensure that we constantly learn from performance through feedback loops. As we like to say “the first piece of AI content you produce is guaranteed to be your worst.”
AI driven analytics also give us the possibility to interpret abstract or qualitative communications measurements and turn those in to real indicators of “What’s moving the needle.”
Q: Do we need anything other than ChatGPT?
See also “We are a Microsoft shop and we are deploying Microsoft Copilot. So we are all covered?” – which addresses the issues of generic productivity tools vs specific applications.
ChatGPT is an amazing and versatile tool that can already transform your workflow. But it does have some limitations.
The biggest challenge is that the average human being is really bad at formulating questions, and even worse at giving specific instructions. This means that whilst chat is a very accessible user interface (one of the reasons Chat GPT has been so successful is its simplicity), it is also perhaps one of the worst for getting actual results.
So we suggest looking at specific applications built specifically for addressing communications & marketing needs with AI.
Q: But will AI take my job??
Probably not – your brain is still incredibly unique and valuable.
See How will AI alter future jobs and roles within communications & marketing?
Q: So it is not a case of Humans VS the machines ?
No. We think there is a more collaborative way of looking at the picture. See What will be the value of human content in an AI-driven world?
AI Applications in Communications and Marketing
Making AI real in your organization
Q: How can AI improve marketing and communications?
There are many AI applications that span the whole communications and marketing value chain.
We could describe dozens of potential use cases, but let’s summarize as: AI lets you properly integrate data into the entire communications value chain (you know how we always talk about being data driven, but really are not). It means that every outcome from planning to execution to amplification and follow up can be based on solid reasoning – using the right toolset.
But, perhaps the biggest benefit is that the first version of any AI output is always the worst. AI’s ability to integrate feedback loops means that it can can constantly improve and optimize, which is one of the reasons why we feel there will be an increasing gap between the AI haves and the have-nots over time.
Q: What are three of the biggest and most obvious pain points in communications and marketing that AI can help with?
From many conversations here are the three major pinpoints we think AI has the most potential to solve:
That a large percentage of today’s content is wasted – meaning that it does not get significant engagement. Estimates vary but often suggest that as much as 80% of content produced falls by the wayside. AI can help here by enabling better analysis and targeting of content.
Related – that people simply don’t have enough time. Enough time to localize and customize content. Enough time to properly research audiences. Enough time to update websites. Enough time for proper social media engagement. And enough time to think about implementing AI! AI can help here by automating away many of today’s tasks.
That today’s communications & marketing technology stack is expensive, over complicated, and doesn’t deliver value: Again, estimates vary, but some suggest that only 20% of marcom tech implemented by the average organization today is used or delivering value. AI can help by providing new and dramatically simpler solutions to old problems.
Q: Can we trust AI ? I’ve heard that it gets things wrong all the time…
See also “I asked Chat GPT a question and it was completely wrong“.
One of the biggest limitations of AI right now is in the way that people use it – usually the same people that like to write articles about the weaknesses of AI in order to make themselves feel better. Technical term: we call this usage mode one-shot adversarial querying.
In many cases users are trying to test or confront an AI to see if it understands something, or if it will make a mistake. The same approach with humans would produce similar, likely worse results.
Consider a different approach, where we work with an AI by giving it feedback and guidance as it works towards the answer. We tell it where it is going wrong and where it is correct. We provide it with a solid, detailed briefing. With a little coaching we can lead it to a better understanding of the problem that enables it to produce more reliable and repeatable output.
This is similar to the way in which we guide teams of humans. More advanced toolsets allow us to work with AI in this way so that we can effectively coach it towards better results for our specific needs. It is all about having the right expectation and the right approach.
Q: What are the most immediate and viable uses for AI in communications and marketing?
This answer is for the 25% of college educated professionals who have apparently not tried ChatGPT yet..
You can start today by giving ChatGPT some bullet points and ask it to create content. Or to provide an existing text and ask for a summary or for a rewrite. With a little skill this can be a huge productivity booster. Or, need an image for a blog post ChatGPT can also help you out there – but be warned, you will end up making dozens of variations.
Obviously to take this further and create long form or on-brand content requires more specialized tools, but the reason that AI has created such buzz is because it is a fundamentally accessible technology.
Q: How can AI help make communications more strategic?
The digitalization of communications and marketing has created a large amount of data – much more than we can feasibly manage. AI applications are great at analyzing large data sets and inferring meaning, turning data into actionable insights. There are applications available that use AI to turn this data into useful output in a fraction of the time taken with traditional methods giving deeper and more usable insights.
Q: How can AI help you to better target audiences?
Beyond making better use of data, another core strength of AI is the ability to quickly create different variations of content. Now imagine that an AI can analyze data to determine the subtle differences in needs between audiences, and then create multiple versions of the same document targeted at each specific audience type. Yes, we can do that. It’s really cool and we think it is an essential advantage in a world where people are bombarded with increasing amounts of tired, generic content.
Q: How can content be differentiated when it is all AI generated?
There is already a clear difference between quality AI generated content, and poor AI generated content.
Off the shelf tools such as ChatGPT can only differentiate their output based on different prompts since they all share the same training data. But there are further opportunities to differentiate. By integrating your third party data into the process, by fine tuning the model to your needs and by setting other advanced parameters it is possible to create content that is much more specific to you and to your target audience. See also above – How can AI help you better target audiences?
We also think there is an additional category of AI content generation that we call “AI accelerated” which provides a higher quality of content that normally produced by humans.
Q: What is AI accelerated content?
This is a word we made up, but it is also a real thing that we do.
We still believe that humans have the edge when it comes to content creation due the subtle extra context and embodiment they are able to provide. AI accelerated content is about combining the best of human creativity with the advanced analytical skill that AI can provide.
The trick is to provide an AI generated blueprint for content which contains research, suggested story arc options, and other rich data that make it easy for the human to quickly assemble a very human story. We can show you more about how we do this.
Q: Will using AI slow down our communications and make us less responsive?
For any new technology there is an initial investment in time to become accustomed. All tools need some familiarization and we would say that AI will certainly speed up (or, we like to say “Supercharge”) communications because:
- The fact that modern AI is natural language based means that applications can be much simpler than other communications tools so that anybody can use them
- AI is able to generate insights and content and do research at a superhuman pace and level
In many use cases AI can produce hours of work in minutes, and in some use cases it can produce months of work in an hour. It does require some familiarization time, but the payoff should be quick.
Q: How will AI-driven language models be used to personalize marketing campaigns and communication with customers?
The early days of the Internet promised us personalized marketing. This certainly happened through the ability to match different profile types with pre-prepared content – but there was still a limited number of content variations based on the cost of creating it.
AI allows us to create completely original, customized, authentic content for each specific individual, quickly and at scale. It almost takes us back to the days of personalized letter writing. This is a huge opportunity to make your communications more authentic.
Q: By using AI will we lose our unique tone of voice and begin to look and sound like everyone else?
A common argument is that generative AI simply takes the average of its training data and spits that out. That is a slightly simplistic analysis.
Whilst it is true that an average “prompt” will give an average answer, when properly fed with data AI can create content that is authentic, engaging and supported by unique tone of voice and containing unique insights.
Done properly AI can be trained on existing content and brand voice to make it sound just like you – even a tiny bit better.
Q: What will be the impact of AI on search and SEO?
A relevant concern is that a huge amount of AI content will flood the internet, making relevant, quality content hard to find. We don’t think this will be too much of a problem – search engines are highly evolved to spot genuinely engaging content that users want to read and separate it from the filler.
This is what we wrote 6 months ago on this subject:
The bigger concern is that our fundamental relationship with search will change. We may no longer search for web pages and then read them. Instead we will be reading a machine generated summary of information from websites, possibly generated by combining results from several sources. The Google’s of the world are already evolving their services in this direction.
In the meantime, Google did this. And it wasn’t pretty.
However, despite this temporary setback it is clear that this is the direction things are evolving towards.
Potentially your website content needs to be designed with this need in mind – does it directly answer the type of questions that people will be asking AI about your business? If you ask an AI chatbot today about your firm do you get good answers? What if you ask it to compare your firm against your competition?
This puts the challenge of creating web content on to a new level. Or actually on to an old level – the content on your website needs to accurately answer the questions that your customers have. But… actually AI can help with that. What if we ask AI to recreate your web content based on popular questions, using your entire product catalogue, customer profile and business strategy as a basis?
Implementing AI – the hard part
Concerns, adoption barriers, challenges – what do you need to know about implementing AI solutions?
Q: Is it worth looking at AI now, or should we wait for the next generation of improvements?
AI is good enough to create value today so there is little downside to adopting already.
At this stage waiting will only widen the gap to those getting a head start and delay the very important process of becoming familiar with an important new technology. Most people would agree that AI will have a clear role in the workplace of the future, so we would advocate for being prepared.
With many companies pushing ahead aggressively the question is more: can you afford to wait? And how hard will it be to close the gap later?
Q: Our company has decided to take a wait and see approach. Is that smart?
See above. We appreciate that caution may be wise when dealing with a new technology, particularly one less proven in the field.
But we would not wait for two additional reasons:
First, it seems clear the genie is not going back in the bottle. AI is here to stay, it delivers real business value and provides real advantages.
But secondly and more importantly AI represents a fundamental structural change that will take time to get to grips with. Waiting only increases the risk of being left behind.
Q: How can we evaluate which AI solution is right for us?
This is an important question considering that over 3000 AI tools were released in 2023.
We suggest starting with the areas of your operation that currently hurt or cost the most, then look at which solutions will fit. Ideally test them out with some actual use cases, make sure that they are flexible enough to handle your needs and then get to work. If you don’t find solutions, look a little further – there will be some.
Probably the most important thing here is to not limit your imagination when selecting the opportunity areas in your organization. Remember, AI can help with everything from collaboration to planning to content production.
Q: What are the barriers to using AI and how can they be overcome?
We think there are three key things – change management, technology capability, and governance, privacy & security.
The change aspects of a technology shift as big as AI should not be underestimated. It will have a significant impact on processes, work distribution and fundamental approach. This is a very complex topic which we have written more about separately.
There is also the human side. Employees are rightly concerned and these concerns do need to be properly addressed – firstly because this is the right thing to do, and secondly because we have seen how change can easily be torpedoed without active employee engagement.
We also must be realistic about technology capability. There are many things that AI is already doing better than humans, but there are still many gaps to be filled and expectations should be kept realistic. In particular, AI often lacks the context that humans can bring to content . This is one of the reasons we heavily advocate for AI Accelerated Content at this stage of AI’s development. .
Then there is governance, privacy and security. Many AI applications don’t fully guarantee this, so make sure that the solution you choose is enterprise grade. Your IT or compliance team will also want to “enter the chat” at this point.
Q: Isn’t AI only really suitable for bigger corporations?
In previous times only the largest firms could afford the kind of skills and technical investment required to implement AI solutions. But today there are many tools available at every price point.
Admittedly some of them are not cheap – but neither are human beings. The business case in terms of productivity and quality of output is in many cases the best business case people have seen for a long time.
Q: How can we integrate AI into our operations today?
Here is a simple process we suggest that could help you become an AI powered organization literally today.
- Make a list of all of your main tasks and activities.
- Put them in to a 2×2 matrix with two axis: “Repetitive” and “Complex”.
- Choose the tasks and activities that high on repetitive and low on complexity.
- Spend a few minutes trying to come up with a Chat GPT prompt to solve that task. (we invite you to email us if you struggle here).
You will probably find that using this process you can find at least a few things where AI can already help today. Just to get you started, here is a draft matrix we jotted down:
If you want to try some of the use cases that are high on complexity then get in touch – we have something for you.
Concerns, issues, misinformation and safety
Things that we need to think about carefully as AI takes an increasingly important role in our workplace
Q: What about safety and concerns about AGI (artificial general intelligence), are the robots coming?
This is a very complex question best answered with the right depth and detail somewhere else.
But for what its worth we don’t see the current generation of AI’s evolving more human like intelligence any time soon. Simply looking at the difference in energy consumption between the human brain and today’s AI models, and the fact that human skills are acquired by a deep understanding of and interaction with the physical world shows that there are still some fundamental challenges to be overcome before we get to human like artificial intelligence.
Q: Will AI cause significantly more misinformation?
AI can contribute to more misinformation because it can easily generate content. However, it also provides us with new tools to better identify and manage false information, making it possible to maintain higher standards of truthfulness and accuracy.
So….
Q: Can AI help solve the challenges of misinformation?
Yes, AI can be very effective in tackling misinformation. It can quickly analyze and detect false information, helping creators focus on producing accurate, high-quality content. This can lead to a more trustworthy information landscape where good information stands out.
The topic of misinformation and content quality are complex issues which we have written about in further detail.
Q: What about errors – doesn’t AI frequently get things wrong?
This can be an issue and a very real risk.
See: Can we trust AI ? I’ve heard that it gets things wrong all the time and I asked ChatGPT a question and it was completely wrong. Can I take AI seriously?
Q: Won’t AI simply lead to more low quality content
While AI can generate low-quality content, it also has the potential to enhance the quality of content overall. By taking over repetitive tasks, AI allows people to spend more time on creative and thoughtful work, resulting in more engaging and well-researched content that benefits everyone.
Q: What regulations should companies be aware of when using AI in marketing and communications?
This varies widely by geography, by industry, and the picture here is changing all the time. For this reason we suggest you speak to your compliance team about that question.
The good news is that if you choose a proper enterprise grade AI solution any issues should be solvable.
Q: Different countries have implemented or will implement restrictive regulations on the use of AI, could this impact us too if we go ahead with an AI project?
After a lot of initial bluster and concern this topic seems to have calmed down.
Most current AI legislation focuses on specific high risk use cases, such as HR applications, or in areas where privacy may be infringed such as facial recognition. Additional legislation focuses on restrictions around the development of advanced AI models which could pose future risks.
And while there has been much concern (and many articles) about legislation currently in the works (particularly within the EU), it does appear that lawmakers are working hard to make sensible compromises that will offer useful protections whilst not stifling innovation or business productivity.
In summary, we do not expect that legislation will have a negative impact on AI in the communications and marketing space, but it is worth watching closely. We will certainly update this document if we see changes.
Q: What about copyright concerns?
We appreciate the arguments that many artists and creators are making, many of whom see AI as taking and adapting their work without permission. At the same time, all human works are based on prior art and knowledge.
AI is trained on publicly available information in the same way that humans learn by reading and consuming literature, art and media. AI then uses a process to turn its training into new and original works (which do not violate copyright).
Whilst there is much legal discussion about this topic and there are many challenges from existing copyright holders we don’t expect this to be a practical issue going forward, even though the root issues will likely be discussed for a long time. We are going to need to find a way forward that works for everybody.
Q: How should we handle customer data privacy and security when using AI?
This should be handled in the same way as any other application you use (AI really isn’t that different) – you need to make sure that the provider can comply with your usual information security and compliance requirements.
Q: Are there specific cybersecurity risks when dealing with AI?
When using publicly hosted or free services there is a danger that the information you send (or the prompts that you give) are leaked. Or that they end up as part of the future training data, meaning that a competitor could ask “what did company X ask about last year?”
After initial concerns most providers have put in extensive safeguards against this so the risk is somewhat low.
Also, many solutions operate in fully private environments meaning that there is no chance of data being purposefully leaked. It is important to understand the right single-tenancy solutions, and ways to avoid certain risks like Prompt Injections.
Q: Is there a danger that we lose control of what we are communicating and who to?
A question we are often asked is: can your AI publish directly, or can it interface with our content or social media management system?
Our answer is usually no, because this would be a crazy idea. (although it is possible if you really like)
Our own solutions have been carefully designed around a “human in the loop” concept, based on the idea that AI can do the work but humans need to provide thought and direction, and must run the decision process at every stage. In fact, we think AI gives you more control since you can easily and quickly make new variations and adapt your briefing in a way that takes far too long with manual processes.
To answer the actual question – there is a danger, but good solutions and processes can mitigate it.
The future and the road ahead
Leveraging AI for long term strategic advantage
Q: What are the AI opportunities beyond content creation?
There are still many opportunities within content creation. Right now AI tools excel at images and text creation. Both of these are continuously improving. Video content and infographics are likely to be viable at a decent level of quality soon.
But beyond content creation there are just as many opportunities (one of the problems is that the world got a bit caught up in the term “generative AI” which made everybody focus on creation). The power of AI to manage large amounts of complex data, to infer relationships and to manage complexity means that AI is just as useful (if not more) in strategy, planning, monitoring, analyzing and enabling collaboration through its ability to have awareness of large amounts of data.
Q: What will be the next big thing?
There will undoubtedly be a stream of smart, interesting and valuable innovations in the near future and the next major generation of AI models will likely be shockingly good.
But we think that the real change will be in the shift from models and technologies (essentially AI training) to their ability to provide business value (what we call Inference). This is the missing link today and the next big thing: the ability to provide seamless business value.
Q: How will AI alter future jobs and roles within communications & marketing?
We think it would be naive to assume that everything will stay the same. But we also wouldn’t panic.
We can look at a historical precedent – social media. Over the past 15 years social media had a dramatic impact on the communications & marketing industry. Today you will find multiple job roles and titles that did not previously exist. At the same time team structures and budgets have also changed, sometimes dramatically, to accommodate the importance of social media.
In the same way we believe that AI will have a significant impact with its ability to automate and reshape value chains.
However, we also think the concept of everybody being replaced by AI is far fetched. The real benefit of AI is that it can make humans more productive and raise the level of output. It can free us from tedious tasks and give us back the time we need to do the things that only humans can do best.
No, this does not mean that we can ignore it. Our favorite saying is that “you won’t be replaced by an AI, but you may very well be replaced by a human using AI”.
Q: What will be the value of human content in an AI-driven world?
Let’s first get rid of a common myth “AI cannot be creative, so humans are safe.”
AI can absolutely be creative. In fact, it can take your basic ideas and apply a level of creativity that brings out the best in them. But where do those original ideas come from?
This is where we see the lasting advantage of humans – the ability to draw on a much broader spectrum of insights, observations, experience and other senses to synthesize original ideas that AI can then build on.
AI is kind of trapped in its own small box, with only the data that we give it. As powerful as it will become, and the more diverse data sources we can give it the better it will be. But this will still be quite limited compared to the broad experience and understanding of humans and their environment.
Humans will still be the boss.
Q: How will this technology develop and what will its impact be in 1, 2, 5, 10 years?
Note, we have been remarkably consistent in not having to update this section for the last few editions.
This is very hard to say, technology predictions are fraught with problems. As Bill Gates once said “we tend to overestimate what can be done in a year, but greatly underestimate what can be done in 10 years”.
That said, we’ll have a go.
Over the next year we are will see incremental improvements to the current AI models, and maybe some slightly bigger improvements, but most of all we will see the beginnings of real mainstream adoption within businesses.
Within the next 5 years we will see some significant improvements leading to human level AI performance across a wide range of tasks. This will enable seamless applications and simple replication of existing business processes using AI.
Ten years is hard to predict, but we expect that by then AI would have made a fundamental change to the way that we do business and the way in which our work is structured. AI forerunners will have transformed their operations and created a wide lead to those who have struggled with implementation and change. Meanwhile we will all be wondering how we ever used to do all that work just by ourselves.
—
Did you make it this far? Thanks for reading! If you have a comment, or another question, or want to get in touch for any reason then please connect with us.