There is no doubt about it, Generative AI is changing things. It is changing how we work, how things look, and how we approach things. We’ve got this new tool that is helping us be more efficient, express ourselves creatively and is fun!
I think it is great. I think it is great to the extent that I get made fun of in the office because I’ve made a gang of imaginary AI friends that live in our chat office and on Instagram.
However, it has made it incredibly easy to be average. I like asking Chat GPT to generate a brand for me and then having a laugh at how generic it is—but it is passable. It’ll lack the depth and insights a skilled professional would bring, but it’ll do the job for some clients. It’ll be average.
Anyone could create it now—for free. Why get a professional involved?
Generative AI, whilst impressive, can have this inherent lack of authenticity and humanness about it. I’ve seen some awful applications that are just off putting. Here are a few examples and how to avoid them:
Completely AI generated content: articles, blogs and comms
AI has made it so quick and easy to create content. Anybody can generate an article about anything within a few minutes—and they are. The internet has exploded with a torrent of this generic, artificial, border-line spam. I hate it. It makes googling for quality information so hard!
When generating entire articles you are not saying anything new, you are not delivering your unique value to your readers and you are not demonstrating your expertise. You are saying that writing something new is too hard and time consuming for you and your readers are not worth your time.
“But nobody will notice”, you tell yourself that, but people are smarter than you think. AI generated articles tend to be repetitive, verbose and very general in nature. Vague and general content is annoying at the best of times.
It also can be wrong. Hallucinations (as it is called) are a real problem with Generative AI, and unless you are proofing work on a topic on which you are already knowledgeable, you could publish false information without realising it.
You will not earn an audience from a stream of Ai generated junk. It lacks authenticity and your personality.
There are AI models that are getting better, getting more nuanced, more refined and more human-like. I might be wrong in time, but people like connecting with people which is something Generative AI can’t do (without making it up).
Putting yourself into the stories in the content is what someone connects with. Anybody can tell 3 reasons why eating too much sugar is bad, but only you can tell it in a way that you would.
People seek connection with other people. Generative AI won’t be able to do this in a genuine way.
What to do instead?
I love to use Chat GPT to refine what I write, or expand on something I’m stuck on. I use it to augment my work rather than replace it. It takes me a lot of effort to keep my paragraphs short and snappy, Chat GPT is my personal editor. It’s created a lot of efficiency for me and improved my work.
Generative AI is best when you as a reader don’t realise it—which is very achievable. This is what you should be aiming for. You wouldn’t know where I’ve refined this article with Chat GPT.
Additionally, feeding Chat GPT or other Generative AI models your completed comms and then getting it to create other applications of it i.e. social posts (which you then review and edit) is a great way to streamline your workflow and scale your effort.
It’s too easy to create generic junk now and too many people are doing it. You are the opportunity.
Blindly using AI generated images
I love making AI generate random images. I’m a crap drawer with a wild imagination. Generative AI was made for someone like me! Getting AI to visualise my crazy ideas is satisfying—from an ice-cream tiger, or a buff unicorn holding a gun or me fighting spam with a big sword in the cyber realm: an AI can generate it basically instantly. We’ve even got an Instagram account dedicated to it!
However, most of the time the results, whilst amusing, look like an AI did it. The images are glitchy, fingers are weird, something isn’t quite right. The latest generative video from Sora by Open AI is incredible and looks almost real, but there is something not quite right about.
DALL.E 3 and Stable Diffusion are making some incredibly real and impressive images, but my LinkedIn feed is not an indicator of this.
How to do it better?
Maybe proof your images first and count the number of fingers?
Seriously though, I’m not sure of the solution here. I understand the appeal. AI generated images are a great, cheap (free?) way of making images for content that you would otherwise have to put a lot of time and effort and money into.
I know the first step is definitely knowing the limitations and weaknesses—detail—fingers and hands in particular.
It is a good solution to the image problem, it just can look naff—which makes you think if it is a good solution? There are other tools that you can load your brand assets into and use it to create images, but that is a few too many steps for most people.
I’d personally be opting for generic unsplash photos over AI at this stage. It doesn’t feel as inauthentic as AI generated images. Some nice photos of yourself at a desk would probably be more effective if you have to have an image—think on my previous point, people connect with people.
If unsplash/stock photography is not your vibe, using AI to create stylised images that lack detail can look really nice. OpenAI do this on their site, they use a lot of colour gradients, or a painting style that uses thick brushstrokes rather than fine art. This is playing into AI’s strengths.
We at BEECH have wrestled with what to do for our new latest section. We’ve opted to just go with a generic image (stock). We felt it is more inline with our brand and site that way.
Do what works for you, be be intentional about it.
Disempowered unintelligent chatbots
This is one I haven’t seen as often in my web travels, but it annoys me the most. A company offers chat as a way of connecting with them. Fantastic! The chat is just a bot, that isn’t smart, that just answers in vague FAQ links that you’ve already read through and has an inability to pass you onto a person.
In this instance, I would prefer not to talk to a real person. No connection required. But, I have a job that I want done and this chatbot thing is making it harder and a person is the only way forward.
Don’t do this. It isn’t helpful. Turn it off. Kill it. It just frustrates your customers/clients. I think waiting on a response to an email would be less frustrating. Or if you do give the thing the ability to look at the pages you’ve navigated to prior so it doesn’t recommend the same things every. Single. TIME.
How else to approach it
The technology in this use case is very emergent. We’re in a transition stage from a chat bot replying with a pre-written response based on keywords, to an generative AI that could offer some semblance of understanding.
The technology is evolving rapidly and a good solution would be transformative for organisations with large customer service volumes—both for the organisation and customer.
If I could chat with a Bot (chat or voice [yep that is a thing]) that could understand and was empowered to solve my problem I would talk to them 100% of the time.
Imagine calling up a call centre and there was no wait, the agent knew the answer to all your questions, was clear in their responses and did the data entry correctly. You’d then believe in miracles right?
I want that.
This is actually possible with the technology we have today. I’m just not sure if a platform exists yet that brings it all together. It wouldn’t have a free tier and you would have to take time implementing it, but it would actually enhance your customers’ experience with you.
At the moment those chat Bots suck and are not helping anyone.
—
Landing the plane
I asked ChatGPT to proof this article and suggest a conclusion and I felt bad. I’ve been rather critical, but I have such big expectations and excitement for the future of AI.
When prompted well and worked with, generative AI can help create great quality stuff. And this is not limited to just text, for instance the tools rolling out in Adobe Photoshop is nuts (albeit a little rogue sometimes)!
When using generative AI be clever about it. Don’t just jump in and start using generated images or copy because it is cool or efficient to. Do it because it will help you bring better value to your customers, or communicate your idea more effectively. Use it authentically.
Back to answer my original question: is AI revolutionising creativity or diluting authenticity?
Yes.
To both.
I think we’re in for an explosion in creativity over the next few years. We’re also going to see a lot of crap.
It’s going to be fun.
