Iterating fast – lots of ideas can be generated, such as a concept in lots of different art directions, shot types or modifying the contents such as substrates, lighting and colour ways.

New ideas - and abstract concepts that we wouldn’t have thought of ourselves. AI is amazing at cross pollination and lateral thinking and merging unrelated concepts to create new ones.

Pre-visualisation – not just new ideas but taking something that would have required a set build, shoot, 3D, VFX or stock sourcing and making it visible enough to discuss.

Filling in gaps - such as linking frames, extending sequences or generating targeted visuals that do not exist in stock.

Generating content in different sizes and formats and automating the process.

Creating targeted and bespoke content across different languages and cultures.

Accuracy - Replicating branded assets such as logos, brand colours, substrates and text.

Consistency - It can create a great product image once. Getting it to reliably create the same product, person, environment or visual language repeatedly is a different problem.

Being a replacement for stock – particularly for video, AI is not a complete replacement, especially when dealing with niche content that requires accuracy i.e. the sciences, models etc.

Revision tax is high – iterative loops can become out of control quickly and push a project out of scope surprisingly fast.

Workflow fragmentation – Too many AI tools and switching between traditional and AI platforms can slow efficiency and create burnout.

Style over substance – it’s intelligent, but it’s not intuitive. It can generate technically impressive and superficially beautiful imagery, but the details could be way off the mark for the brief.

Assuming AI can make it faster – it mostly wont. Which is why you should focus on effectiveness, instead of efficiency. Unless you have a framework in place to make that happen.

Thinking AI’s capabilities are greater than they are – it still isn’t a replacement for most skill sets because it’s probabilistic, not precise.

Substituting large volumes of AI generated imagery and creating ‘options’ instead of creating a strategy to focus the decision making. Choice isn’t the same thing as creativity.

Thinking the prompt is the problem – You can spend ages tweaking prompts, when it’s really the creative direction and/or strategy.

Regenerating when they should intervene – trying to solve every problem with another generation instead of making a decision.

Leadership will assume a lot of things about AI which won’t be true. Working at the forefront of AI and design requires delivery + education about its pros, cons and limitations.

I have spent a lot of time in the weeds figuring out the ‘how, when and why’. There’s now two processes – designing the creative process and simultaneously delivering it. That can be tough.

AI burnout is real – the mental drain that can be created by constant iterations because the AI just isn’t precise enough can be acutely draining. I almost quit a project once because of this.

Having a framework around how to use AI in a creative project is now an essential, not a nice to have and it’ll become more important as time goes on.

IP, regulations and compliance are important but there can be a large disconnect between the end users designing and compliance functions in global orgs. There needs a more direct bridge between the two.

The bigger the AI perception gap is, the more unrealistic the client’s expectations will be. It’s not deliberate, but because they won’t know any different. Ask the client or team upfront their understanding of AI tools first and see what knowledge gaps you may need to fill.

Assess the tools for the job. Can we create what’s in scope for a project in time, budget and within the constraints of the approved AI tools within that particular organisation? Are they logistically prepared to produce at scale?

The cost of AI is real. Creating AI content at scale can be surprisingly expensive as everything is credit based. Many organisations are underprepared for just how much that’s going to be, and you’re likely going to be the one to tell them.

Navigating the AI ecosystem inside the whole organisation is now part of the job. AI is part of the inherent infrastructure which spans compliance, IT, licensing, security, budgets, approvals, brand governance, and IP.

You become the person who has to explain why AI can’t do something. Leadership hears ‘AI can make anything possible’ You’re the one who has to explain "Yes, but not that, in that format, with that level of accuracy, within budget, using those approved tools, by Friday."

Understanding the brief - It doesn't understand the brief in the way a human does. It can interpret instructions and patterns, but it doesn't have an inherent understanding of the strategy behind the work. That is a very important distinction.

The idea – What are we trying to say and why are we saying it? How are we relating this to current culture, is it relevant and does it  feel like us?

Quality control – this is a big one. At a glance your content may look perfect but hallucinations, continuity mistakes and other errors can occur and they can be unnoticeable…at first. Knowing what can go wrong with AI gens becomes a valuable skill.

What deserves to exist – Out of the options, what deserves to exist and why. Someone needs to be able to say ‘this one and here’s the reasons why’.

The ethics – There’s still discussions around the ethics of how we’re applying AI to certain projects and ethical implications of those, there is often discourse between the agency/brand, client and legal.