Happy Birthday ChatGPT! ????

I get asked a lot of questions about AI at work. Most of the questions could be answered by ChatGPT but people still come to me. This made me realise that there was a bit of a gap between what laypeople perceive AI to be capable of, and what it is really likely to deliver for many of us, especially those not working in tech, where it's already pervasive.

When people ask me about AI, they are almost exclusively asking about ChatGPT* - this thing they have heard of but are not sure how to use - and there is also the occasional questions about the existential threat to humanity. As a self-proclaimed non-expert, I'm most interested in thinking about what LLMs and AI chatbots can bring to the workflow and productivity of everyday business people, and what trends we might see in the near future.

So on ChatGPT's 1st birthday, I thought it might be fun to summarize some of the things I have found these tools can be used for, that are within reach of anyone, even if you don't know what AI stands for...*

*Alternative LLM-chatbot-interfaces are available. You might need to pay for ChatGPT premium to get some of these features, but this is always changing and is expected to remain fairly inexpensive. There are also plenty of other AI tools out there that can do the same, or better, for free in some cases. I will refer to ChatGPT mostly, but all this advice applies to plenty of other competing products out there.

What can anyone get out of these tools today?

Example #1: Collating notes

Starting simple, the most obvious time-saver for the everyday user is reading and summarising text. And that means handwriting too.

Got a load of post-it notes you need to type up after a big workshop? AI can help with that. Handwritten and hard to read? AI can also have a go at that. Need to also sort them into groups of similar themes, like positive vs negative? AI can also do that pretty well. Need it in a format you can paste directly into Excel? Also no problem. Just upload a photo of what you need processed and ask for the output you want.

The only caveats here are (1) obviously don't upload confidential or sensitive information anywhere you shouldn't be and (2) check the results, especially if you uploaded handwritten notes.

The final thing to remember is that this is not about getting polished meeting notes out of an AI tool, as that can still be difficult and time consuming to get right. But if it speeds up the menial parts of the task, allowing you to get on with the interesting, value-adding work, you're winning

Example #2: Generating images

The header image for this article was generated asking ChatGPT for something suitable for linkedin, that was 1920x1080 pixels and summarised my theme of chatbots in the workplace. I didn't like the first image it generated so asked it to try again, simplifying the style. Hey presto, with AI's help it turns out 'second time's a charm'.

I've seen this used to great effect when trying to generate professional-looking images to take a presentation to the next level, where there is no budget for formal graphic design help.

Example #3: Getting fast feedback

Ever been stuck waiting for feedback on a first draft, only to realise that you've actually created a bottleneck by relying on someone who is too busy to help? Why not bounce your ideas off a virtual teammate for a bit first, and be able to share a more considered 'v2' document instead, with (hopefully) far less input required to get it over the line?

ChatGPT has been used extensively by people to 'act as' a certain type of colleague, customer or consumer which can help you get interesting and new feedback that you might not have thought about, to accelerate an iterative drafting process. Developing a new product idea? Ask the AI to act as a consumer and ask questions or poke holes in your idea. Reaching out to a tough new customer? Ask the AI to act like a dis-engaged prospect in the relevant sector and pre-empt some of the challenges you might face.

Example #4: Writing executive summaries

This one may not always be possible due to data confidentiality, but if you've ever struggled to put the final introductory paragraph together for a large document or other deliverable, try asking AI to read the document and giving it a shot. The benefit of this is you can combine it with example #3 above and ask the AI to inhabit certain personas as it reads, prioritising different aspects of your content in the summary. You can also stipulate how long the summary should be, the tone, whether to include facts and figures or not etc.

The summary will likely need tweaking but it is a helpful way to get a truly 'fresh pair of eyes' looking at a long document and will even help you develop different summaries to be used with different audiences.

Example #5: Brainstorming anything

Need a good pun for a LinkedIn article title? Ask AI for a few ideas, and even if they're bad, they're guaranteed to minimise any writers bloc holding you back. Need to come up with a catchy acronym for a new process you have developed? Ask AI for ideas on how you can creatively put it all together. Need a good tagline for a social media post that is more likely to engage people? Get some initial ideas (including emojis!) from an AI.

Again, this is not about getting AI to do all the work for you, but just speed you up at the tricky stages of thinking and content generation

What should we expect to see more of in the future?

This is where it gets a bit more interesting for me, as many people may have heard versions of the tips above, and more, from LinkedIn and Twitter already. People always enjoy trying to predict the future, but I think with AI we have a tendency to get carried away and I'd like to be a little more realistic.

Buy (not build)

2023 has been dubbed the year of AI by many, understandably so. But many organisations (especailly those that are not tech-companies or startups etc) are only just getting to grips with how to harness this new capability. I think this means uptake will still be quite slow in many traditional sectors, and of course it will be outsourced to experts, especially the big players, i.e. Microsoft.

We have already seen announcements of Microsoft's Copilot which will be embedded within the suite of tools many professionals use every day.

AI-readability

I've spent a lot of time creating analysis, dashboards and presentations in my career. All of these have been for human-consumption. I do wonder in the future if there will be times when information will still need to be structured by humans but the primary consideration will be to ensure AI can consume the information as efficiently and accurately as possible.

Why should we spend time creating visuals that allow humans to understand a story, when in the future they will just be asking an AI assistant their core question and receiving a succinct answer? In the future we actually might get a better return on our investment by writing documents with specific AI-readability in mind. This will be adjacent to the search engine optimisation skills many digital marketers focus on today.

CEO interactions

Following on from above, in a world where leaders and managers can get more and more of their information from an AI assistant, how will management structures and meeting routines start to change?

I think the big thing we should see is a reduction in 'decision by committee' style meetings. There will be less debate over what particular business issue needs to be solved, and how to solve it. But there will still be a need for humans to converse in small groups to take physical actions in the real world, and motivate each other and hold people to account.

Digital team members

I often like to think of ChatGPT as a person who I can bounce ideas off. This is especially useful when asking it to 'act like' a certain demographic of customer, or role in a business, for relevant feedback.

However, as the technology advances, I expect that the need to ask the AI to act a certain way will disappear. Instead, hyper-specialised AI assistants will be 'bolted-on' to teams in order to increase capacity and throughput, but still plug-in to the human-based, conversational nature of work. (Eventually this might become more seamless, but I am thinking about the next 12-18 months).

These digital team members will retain knowledge of the specific business, team and activities performed, and will learn quickly. They might be expensive, if they work well, but could be switched on and off in sync with the ebb and flow of the business.

We've seen the start of this with Open AI's GPTs (specialised versions of ChatGPT) and I think this model will catch-on inside businesses.

A more sophisticated cyber security arena

Plenty of us have been there - you've just taken the latest cyber security training at work and are very confident you'd never be fooled, then you get caught out by something completely new (and you cross your fingers hoping it's a test email from the IT department).

Things are only going to get tricker in this area. AI tools that can write like humans, generate new ideas and personalised responses, can generate human speech, or imitate a human voice, will all be used to fool us.

The multiplier here is that tools such as ElevenLabs - that can listen to your voice and then produce a text-to-speech product that sounds like you - are going to pose the biggest threat to people who have a lot of recordings of their voice out there on the public internet. And guess what, these are the CEOs, Celebrities and Influencers of the world, all who have great sway over certain groups of people.

Never mind the phishing emails of old claiming to be your boss, wait for the fake 'angry CEO phone call' and see how confidently you can question their veracity for fear of retribution...

Closing thoughts

I just used ChatGPT to spell check this article and then suggest any minor improvements (but not wholesale changes, becasue I didn't have time) and it did a better job than my browser at spotting mistakes. It also suggested I add a conclusion, so here goes:

• AI tools can already do some very simple and useful things - start using it where you can, even if (especailly if!) you're not a technophile
• Don't get distracted by people talking about what AI will do in the long-run, because it's impossible to predict - start thinking about what will change in 2024 if people start doing simple things differently
• I think the impact on management, leadership and meetings is more important to many people working in traditional businesses than concerns around job losses and 'Skynet'