On 30 November 2023, ChatGPT will celebrate its first birthday. Initially seen as a curiosity and a novelty, the chatbot is now firmly part of the daily lives of most marketers.
But, while ChatGPT has undoubtedly brought AI technology to the masses, it has also created a host of concerns, including worries about job losses, the spreading of disinformation and accusations of plagiarism. So, are we better off than we were 12 months ago? And what does the future hold for AI? Let’s take a detailed look.
How has the use of ChatGPT changed over the past year?
When it was first released in November 2022, ChatGPT from OpenAI was greeted with a huge degree of scepticism. In its infancy, it was primarily used for novelty purposes, with many users asking it to write songs, answer riddles or play trivia games with them.
However, as ChatGPT proved its accuracy over time (and withstood a barrage of complaints and ridicule), the focus of users changed. Now, it’s reliably used for a great range of tasks, including drafting emails from scratch, penning cover letters and even creating recipes.
With this in mind, here are five handy ways that people have started to use ChatGPT since it first launched. All of these were unimaginable just 12 short months ago.
- An acronym, jargon and slang decoder
Are you a marketer that knows your CRO from your PPC? How about your GAAP operating income from your EBITDA?
Well, if you don’t, ChatGPT is your friend. After all, since the chatbot launched late last year, an increasing number of people have used it as a jargon decoder.
But, while some people may have simply used ChatGPT to either decipher some medical terminology or work out some new marketing acronyms, others have taken it a step further and used the chatbot to craft a contract or create a cover letter.
On top of this, ChatGPT has also been used by others as a form of intermediary. This is because the chatbot can be used to translate workplace jargon so that everyone’s on the same page. This is particularly useful for trans-atlantic companies with offices in both the UK and the USA, or companies that have large generation gaps between employees – think of ChatGPT as your very own Gen Z translator.
- A critic, feedback provider and tone alterer
Looking for feedback on your pitch doc, your design or your copy but think your co-workers and family members are being too kind? Try using ChatGPT.
Many copywriters in particular have started to use ChatGPT to audit their work to see whether they’ve either hit the brief or are meeting the expectations of their audience. Somewhat shockingly, the chatbot can even critique works of fiction and point out flaws like missing characterisation or weak imagery.
Of course, there are ethical questions that are linked to using ChatGPT to review your work. After all, if it rewrites sections, is it still your work? This is a particularly important discussion point in schools and universities across the land. But, while some argue it can constitute plagiarism, others believe that by asking it to be your critic, and learning from its feedback, you can improve your writing without letting it put words in your mouth (or write it for you).
In the marketing world, using ChatGPT as a critic and feedback provider seems like a no brainer. That said, to get the best out of the technology, you need to be clear with exactly what you’d like it to provide you with feedback on. A prompt like ‘give me feedback’ is likely to be much less useful than something like ‘I’m writing this ecommerce landing page. Please tell me if it’s persuasive, well structured and clear’.
At the opposite end of the spectrum, you can also use ChatGPT to tone down your responses to certain problems in the workplace. Particularly annoyed by a work email but concerned you may seem a bit negative or snappy in your reply? Simply provide ChatGPT with your email and ask it to make you sound a bit nicer.
- A whiteboard interpreter
Earlier this year, OpenAI released a ChatGPT update that allowed the tool to ‘see’. Ever since, many users have been impressed at the chatbot’s ability to interpret images.
In a now semi-viral clip, an AI developer called Mckay Wrigley shows ChatGPT a hand-drawn flowchart on a whiteboard. It then turns the image into code that Wrigley runs – and it works. The platform can even tell that Wrigley’s green arrows indicate the steps need to be reordered.
This is particularly great news if you never learnt to code or find that whole process far too manual and laborious. Now, all you need to do is draw a flowchart and let ChatGPT do the rest.
- A summarising tool
If you’re a marketer who struggles dealing with large quantities of information, then you can use ChatGPT to condense it all into bite-size chunks.
So, whether you don’t have time to read a long blog post about outreach, can’t quite bring yourself to read a full chapter of that marketing book you’ve been meaning to read, or don’t want to listen to a full TED Talk about growing your agency (you’ll need a browser plugin for this one), it’s time to turn to ChatGPT. It’ll just give you the highlights, so you can digest that information and then decide if the rest is worth reading or watching.
- A recipe creator
Okay, this final one doesn’t apply to marketing. Well, unless you’re a committed marketer who often works late and forgets to plan their lunches and evening meals (we know there’s plenty of us out there).
So, it gets to dinnertime and you finally switch off your laptop and head to the kitchen. But, although you have loads of ingredients in the fridge and the cupboard, you can’t think of what you can cook. Sound familiar?
Well, many users have reported that ChatGPT has been incredibly useful for creating recipes based on what they have available at that moment. Plus, OpenAI, the developer of ChatGPT, recently added an image-recognition feature that makes this task even easier. Now, instead of having to list the ingredients you can have, you just need to snap a photo and ChatGPT will create a recipe for you.
However, we should stress that with this one, what’s flavoursome and tasty to an AI chatbot may not be up your street. So, although lots of good reports have come flooding in, the technology isn’t failsafe.
We’re not sure a ‘carrot and grape salad with muesli crunch’ recipe that one Guardian reporter received sounds particularly appetising. Added to this, there have been some concerning reports that ChatGPT may be asking some inexperienced chefs to undercook meat. So, if you’re unsure, check cooking times with a reputable source.
A word of warning…
As we’ve alluded to, ChatGPT’s capabilities have improved markedly over the last year. However, although the technology is incredibly smart and helpful, it’s important that users remember it remains an AI chatbot that’s still learning and developing.
For example, although it can recommend recipes, it cannot taste food. Similarly, although it can advise you on email and copy drafts, it does not feel emotion in the same way we do. As a result, its instructions shouldn’t be followed blindly.
Added to this, it can still be downright stupid at times and the internet is full of ‘epic ChatGPT fails’ and examples of the technology providing incorrect or downright stupid answers.
After all, in the past few months, ChatGPT has struggled to count the number of N’s in ‘banana’, failed to correctly answer a riddle it asked a user and agreed that 1+0.9 makes 1.8.
Far more dangerously, it allegedly makes up ‘facts’ – such as a sexual harassment scandal that didn’t happen, starring a real professor, according to the Guardian. Others also remain concerned about where it sources certain pieces of data from and whether it’s spreading disinformation or ‘fake news’.
What does the future hold for ChatGPT?
It’s clear to see that ChatGPT can provide us with a lot more assistance than it could when it launched only 12 months ago. However, with so much development and improvement in only a year, it begs the question: what will the next 12 months look like?
Well, almost exactly a year after the launch of ChatGPT, the product has been given its biggest upgrade to-date. In fact, for ChatGPT Plus users (who pay $20 per month), the changes are almost revolutionary.
After all, the Plus interface has a new look and features. Added to this, Plus users can now create their own GPT bots. On top of this, OpenAI has finally combined most of the extra GPT-4 features into its most powerful model, meaning users now don’t have to select each one from a dropdown list.
These significant upgrades were announced at OpenAI’s first developer conference, called DevDay, on 6 November. The company also released a blog post detailing ‘dozens of new additions and improvements’.
Let’s take a look at four of the most significant announcements in greater detail:
1.The ability for subscribers to make their own custom AI chatbot
ChatGPT was once a ‘one-size-fits-all AI sidekick’, but this latest batch of updates means the chatbot is now highly customisable.
Now, ChatGPT Plus subscribers will see a ‘Create a GPT’ option on their interface. This allows them to create a version of ChatGPT that’s tailored to their needs. For example, a business could make a chatbot that helps their employees, or parents could create chatbots that would help their kids complete their homework.
The good news for non-coders is that the process of creating a ‘GPT’ is straightforward, even if there are a lot of individual steps involved. Users are asked to complete a ‘GPT Builder’ quiz that asks questions about the capabilities the chatbot needs to have, as well as the name and the logo they’d like. Users can also upload data for the chatbot to use as the basis for its responses.
2. An ‘App Store’ for chatbots
OpenAI has also announced a new GPT Store, which will soon be home to thousands of custom AI chatbot creations.
This App Store will be searchable and it will feature GPTs on a leaderboard. However, it won’t be a complete free-for-all. For now, the store will only feature GPTs created by ‘verified builders’.
At present, it already contains over 200 customer GPTs and a new one is supposedly added every three minutes. OpenAI says that it will soon spotlight the most useful and delightful GPTs it comes across in categories like productivity, education, and just for fun.
OpenAI has also stressed that, in the coming months, developers will also be able to earn money based on the number of people using their GPTs. However, the company has not revealed how much of a cut it will take from any GPT sales (for context, Apple takes a 30% fee from app sales).
3. A new GPT-4 Turbo model
Back in March, OpenAI lauded the release of a new ‘multi-modal’ GPT-4 model that allowed people to input things like speech and images into the chatbot, rather than just text.
At the time, this was an important step in the evolution of ChatGPT. But now, OpenAI is going one step further and is boosting GPT-4 (and GPT-3.5) by releasing new ‘Turbo’ versions.
The main benefit of the new GPT-4 Turbo model is that it has been trained on data up to April 2023. This is a vast improvement on the previous version, which has historically struggled to answer questions about anything that has happened since September 2021.
Added to this, the ‘Turbo’ model will also improve the quality of ChatGPT’s answers in long conversations. Plus subscribers can now see a ‘conversations. Plus subscribers can now see a ‘context window’ that refers to the AI chatbot’s ‘memory’ during text chats, which allows it to apply context to its answers.
Put simply, the larger the context window, the better. GPT-4 Turbo can now handle the equivalent of 300 pages of text in conversations before it starts to lose its memory (a big boost on the 3,000 words of earlier versions).
Lastly, OpenAI has also made ChatGPT easier to use. This is because users will no longer need to switch between different models – you can now access DALL-E, browsing, and data analysis all without switching.
Why does this matter?
The launch and evolution of ChatGPT is important for numerous reasons. Firstly, ChatGPT is almost single-handedly responsible for transforming artificial intelligence from a futuristic concept to a daily reality.
Secondly, since it debuted in November 2022, the chatbot has been adopted by tens of millions of users and companies who have been investigating how ChatGPT can make their lives that little bit easier. This includes a number of marketers who are using the chatbot to automate some of their more manual tasks and increase their productivity.
Crucially though, while ChatGPT has proved its worth by creating code, summarising documents and deciphering jargon, it has also done something much more impactful: it has set off a race among big tech companies to launch their own offerings based on generative AI.
Now, Google has imminent plans to launch its answer to ChatGPT, which is called Gemini. Meanwhile, Amazon recently announced that it is investing up to $4bn in the AI startup Anthropic. Similarly, Meta has largely abandoned its Metaverse plans and has begun to pivot its efforts towards generative AI.
Due to this, within the next 12 months, it’s highly likely that ChatGPT (and each company’s own offering) will be even bigger, better and more impactful. After all, in an arms race between companies with seemingly limitless pots of money, we’re likely to see new features unveiled incredibly quickly.
Due to this, there’s already some hope that AI technology will soon be able to help complete specialist tasks, such as helping GPs coordinate healthcare, helping judges summarise cases and helping prevent bioweapons attacks – it appears as though almost nothing is off the table.
That said, at this stage of early development, it’s also important to stress that the downsides of AI are very much real (especially if you recently tried to cook one of the chatbot’s recipes that advises you to undercook your chicken). But, on top of fears over accuracy, many remain concerned about how many jobs AI will take from the labour market, how it will reshape life as we know it and how it currently attributes its findings. Plus, as well as concerns over plagiarism, there are also unknowns about how the tool could be used by children and students across the world.
For now, these wider concerns are for lawmakers and governing bodies to discuss. However, it’s clear that while they were initially thought as issues that could arise in the future, these problems are actually here right now (although it’s hoped that many are purely speculative).
One thing is without doubt though: progress in the field of AI has been incredibly swift over the last 12 months. So, even though AI isn’t perfect and valid concerns must be addressed, we can’t wait to see how the technology will evolve further over the next year.
The latest features announced by OpenAI give us a great insight into the future of the space and the direction of travel. Over the next 12 months, it looks like AI will become much more personalised and customisable, as well as more powerful and detailed than ever before. Through customisation and personalisation, it appears as though these chatbots will allow users to accomplish very complex problems very quickly.
Plus, it looks as though third-party developers will soon be able to create their own GPTs and sell them exclusively through the GPT Store (much like Apple’s original business plan with its App Store).
As a result, over the next 12 months, we expect to see ChatGPT as well as generative or conversational AI get launched further into the mainstream. After all, the GPT Store idea could help users achieve much more with the help of AI. On top of this, they’ll also presumably be able to tune those tools to their own needs (for example adding their own data set). Ultimately, this functionality will likely take AI far beyond the vast, generic abilities that come with ChatGPT today.
And finally, while OpenAI remains locked in an arms race with the likes of Apple and Meta, it’s important to stress that the GPT Store idea could be central to turning the business from one with a broad computing foundation into a business that lots of people pay to use. It could provide the company with its own ‘iPhone moment’ and leave its rivals scurrying to catch up.