It’s the second week back of Jan and you’re trying to get serious about AI in your business. Look no further…
There can be no doubt about the potential impact of large language AI on the media, marketing, business and every person on this planet. McKinsey is estimating the benefit to global businesses in the trillions of dollars. With the automation and integration of AI into business processes, especially in the knowledge economy. There is fairly wild talk of mass redundancies and software costs falling by at least 10 times (don’t believe the hype…). But how does this translate to the average person in the marketing industry?
Let’s start with the basics.
Experiment
The best thing you can do now, if you haven’t already, is to experiment. Try the basics, get a free OpenAI account and start using ChatGPT. When I show this tool to my friends, I get the same reaction I did 20 odd years ago demonstrating how to use the internet — a mix of incredulity, amusement and disbelief — that this technology is going to fundamentally change the world!
100 million users can’t be wrong.
ChatGPT was the fastest growing app or technology in the history of the world. And it’s amazing but also deeply flawed. You’ll start to understand this as you experiment. Ask questions large or small, philosophical or pedantic, technical or general. It has answers to it all. I tend to think of ChatGPT like a ‘know it all’, super pedantic friend rather like Mr Spock from Star Trek. If you don’t ask the right question, you don’t get the right answer. And once it has started to dig its heels in over a question it’s often better to start again. The same is true for Claude, Anthropic’s answer to ChatGPT, it is especially touchy about what it can and can’t do when asked. The better the question, the better the answer.
Once you have tried the free version of both ChatGPT and Claude, start paying the modest monthly subscriptions, the improvements in usability and capability are astounding.
Tools
A poor workman blames his tools…
With AI, as with almost any system, rubbish in equals rubbish out, once you have subscribed and you are able to load lots of data into the AI systems, I cannot stress enough, that the data needs to be of the right quality. That might mean: customer data, spreadsheets, copy, product manuals, financial data, data from the internet, whatever. Always check the output.
In my experience, more than errors or hallucinations, these tools can produce bland, non-specific answers, pretty much like a professor talking about marketing with absolutely no practical experience. You get a general answer that holds for most cases, not the specific instance you might have asked for. I’m getting tired of emails, articles and documents with this bland ‘mush’ of copy.
At this level, use AI as the starting point, not the finished product.
If I’m going to use a tool, what tool should I use? Here’s my starter for ten:
These incredibly powerful tools are more than enough to get you started and contain enough features, such as the ability to ingest spreadsheets, documents, websites, contracts, emails to provide output in many forms, to provide hours of fun. DALL-E, for example, produces top quality illustrations of almost anything you can describe. Most have mobile apps too.
Here are a couple of real examples:
1: I’m not bad at using Excel, but have never mastered Vlookup, Match or Index functions. I wanted to append a few tens of thousands of records with categories based on employee name. I knew this is possible in Excel but decided to let ChatGPT have a go. I loaded two tables into the web version and asked it to append the data. It succeeded after a few prompts, followed by a lot more, very specific instructions. It immediately made me better at this job, more skilled at Excel than I am.
2: We used the new persona ‘skill’ from ChatGPT to load up a LinkedIn profile of a prospective client and asked ChatGPT to assume that persona and then evaluate a proposal we were about to submit, and to think about it from their perspective. It suggested a few weaknesses and improvements in our pitch.
What to do now?
Here’s a handy checklist to get you on your way, tick what you have already done, then work out a plan of action:
- Experiment (subscribe, ask for information, translate, upload files, generate images, create personas, create learning plans, chat with your AI)
- Assess internal adoption in your business, send out a survey to staff to see who is using it and their level of competence and ability
- Assess client adoption, what’s their capability?
- Assess client requirements and appetite for AR based solutions
- Build some tools to help (anywhere in your business)
- Build your own competency
- Don’t forget security, I wouldn’t upload our own or client sensitive data to a public AI
What to avoid:
- Bland application of AI to copy or documents you don’t have the time or interest to do properly
- Assuming AI will revolutionise your business without considerable effort
- Oversharing your private data in public spaces
- Mistaking AI Capabilities for Strategy — AI is a tool, not a strategy
- Failing to Understand Limitations and Contexts – AI is not a one-size-fits-all solution
- Overlooking User Experience and Design Principles, don’t forget the human in the equation
What will the future bring?
There is massive change coming that’s driven by AI, but like all change, it will grow from what’s here already. I have no fear about the existential risk posed by AI, that’s just hype. But I am worried about the bland, thoughtless application of these tools and I worry about increasingly incomprehensible copy, badly designed apps, silly, overly literal, image generation and AI biases based on the training data.
We really can change the world — but let’s get the basics right first.
Featured image: Lukas / Unsplash
