What Can AI Actually Do? Interviews, Marketing & Replacing Salma Hayek
5 in 5 - Brave & Heart HeartBeat #160 ❤️
AI, Artificial Intelligence, Large Language Models – whatever it’s being called, it’s literally all anyone is talking about right now.
Our LinkedIn feed is full of post after post about how to use ChatGPT to become a millionaire overnight, how to get AI to do your job for you, and people claiming to be “Prompt Engineers”. In contrast, news outlets take a much more sceptical, even fearful, approach.
The first wave of hype was simply about the fact that it existed, but now we know what it is, what can it actually do?
Let’s get into it.
Were you forwarded this? Not a subscriber? 👉 Sign up here
#1 - Help You Find A Job
You must have seen them, the posts upon posts about ChatGPT hacks that will make you a millionaire, and the LinkedIn profiles describing themselves as AI experts and “Prompt Engineers” – someone who knows how to give prompts to language models to get the best answers – yes, that’s a job now.
While these posts are honestly largely ignorable, a great concrete example of what the average person can actually use AI Large Language Models for can be found in this article about how you can use AI to help you in a job search.
It notes that one of the best uses of AI chatbots is to help you think a little differently as they make connections that may not have occurred to you. In this vein, you can ask a chatbot what kind of roles are available for your skillset, or for example if you want to change careers you could give the prompt “I’m a teacher, what other jobs can I do with similar skills that aren’t teaching jobs”.
You can use it to help understand the role you’re going for, for example by asking, what does a tech journalist do in their day to day, or what should a project manager know when starting their first job? The bot will regurgitate facts it’s learned from the data it was trained on, but that’s enough to answer your question, pretty much.
It can help you prepare for interviews. Ask what’s the best response when you are asked about your weaknesses in a job interview, or what interview questions are a tech company likely to ask? Again, they won’t be the “right answer”, but it can give you a good starting point to muse on.
You can also ask for a helping hand with cover letters, the bane of everyone’s lives. Ideally you wouldn’t get the bot to write the whole thing, although many people will, but by feeding some information about the role a tailored response will help the brain block that can come with a request for a cover letter.
We do predict, however, that cover letters may soon be scrapped entirely due to the fact that most of them are probably written by chatGPT nowadays.
Don’t lean on it too heavily, otherwise it’ll be chatGPT that got the job, not you…
#2 - Generate Creative Content
ChatGPT works with what is called generative AI – it generates data, in this case text, in response to prompts. As a reminder, it does this by learning the patterns and structure of input training data to generate new data that has similar characteristics.
Can generative AI mimic creativity? In a word, yes. If the input training data contains creative materials, a Large Language Model can generate something with similar characteristics – in this case, creativity.
While the philosophical debate can go a lot further, if what we want to know is can generative AI come up with a fun social media post about a bakery and their amazing cupcakes, it can, with the right prompts and the right training materials.
Content planning platform Plannable has added generative AI to their product, giving four inbuilt “reset” options when you’ve written a post you’re not entirely thrilled with. They are:
Make it Shorter – Does what it says on the tin.
Expand Concept – Takes your idea and runs with it.
Make it Punchier – Fine-tunes the post to, well, make it punchier.
Surprise Me – Rewrites the post in a different tone of voice, and then another, and so on.
You can also give custom instructions, from something simple like “write a post about cupcakes” to “write a post for Facebook with a Christmas themed joke about cupcakes in the style of Richard Shakespeare”, emojis and hashtags added automatically.
We must say, it sounds pretty nifty for when you have writer’s block. And yes, we checked, ChatGPT will do basically the same thing. There is something a little bit weird about the ChatGPT ones, although we can’t quite put our finger on what. But basically, it works.
What generative AI can’t really do, however, is analyse your business, your target audience and your current market and come up with a successful social media strategy. The content, it can do. The thinking behind it, you’ll have to do yourself.
Or, you could always ask us. We know a thing or two on the subject…
#3 - Replace Salma Hayek
We hope by now you’ve all seen the first episode of Black Mirror. If not, spoiler alert, (kind of – you find this out pretty soon). The episode follows a woman who has a Netflix series created about her almost in real time, played by Salma Hayek.
But, it’s not actually Salma Hayek, it’s an AI generated version of her – she sold the rights to her image, and Netflix can make her do whatever they want.
A recent exchange on Twitter shows us what’s already possible. Some Guy (we’ll be renaming him that for this article) posted a video of Ryan Reynolds wearing glasses and talking about how good Tesla cars are.
This video was made entirely by AI, and without the involvement of Tesla itself, or Ryan Reynolds for that matter. Twitter users encouraged Reynolds to sue, he instead made a fake video of Elon Musk touting his gin brand.
All in good fun. However, despite the bad quality and clearly fake people in the videos, the comments were flooded with people noting that THIS is why the writer’s guild are on strike, and the actors guild are following. We’ve all seen how fast this technology is advancing, and their strike request seeks a contract protecting against income loss from the “unregulated use of AI”.
Their fear is exactly what happens to Salma Hayek in Black Mirror, automated versions of them doing whatever producers want them to. Wired magazine gives an example of an actor signing up for one series of a Vampire show, and in season two off they go again, using an AI generated character based on their likeness and performance in season one, with no extra compensation.
#4 - Be A Marketing Buzz Word
Just because it’s AI doesn’t mean it’s new. Recently on our LinkedIn feed someone shared an “exciting” “new” “invention” powered by AI. Yes, we’ve put quotation marks around all of those things, because it is none of them, except powered by AI.
The innovation in question is a feature which is part of a video editing tool that edits the users eyes to ensure that they are making eye contact with the camera. The post went on a bit about being AI-powered, and of course used the AI tag.
Sound familiar? Yeah, that’s because Apple launched the exact same thing integrated in FaceTime with iOS 14, in 2020.
Although it may seem that way when we look at the post because of all the hype surrounding AI, not everything that can be described as “powered by AI” is new.
The auto correct function on your phone has always been powered by AI. It takes inputs from dictionaries but also learns your specific patterns to pick up the words you use most.
These AI models are improving, which is what will allow iOS 17 to learn your preferred swear words, and stop suggesting other words in their place, like duck, for example, but they’ve been in common use for a long time.
So next time you get in a tizzy about the exciting new AI technology in a certain product, ask yourself, is it new, and is it actually exciting? Whether it’s AI powered or not, is not always the point.
#5 - Make The Library Obsolete
One of the main uses everyone has been talking about with ChatGPT is its ability to distill information, and provide information from an open-ended question, unlike search engines. For example, feed ChatGPT an article and it will summarise it. Theoretically, this could help with quicker learning, academic writing, and understanding difficult concepts.
We asked ChatGPT the question “What happened to Marilyn Monroe” and got a few paragraphs about her life and legacy – very useful for a school project. No more scouring the web or, god forbid, books, for information which you then need to craft into paragraphs yourself.
We often compare the fears surrounding ChatGPT and schooling to past reactions to new technologies. Just look at writing, from pen and paper to the typewriter to the keyboard – not all of these advancements went down well with traditionalists, but they seem normal to us now. Back in the good old days students had to peruse the library, now they can find resources at the touch of a button.
However, so far, new technologies only enabled us to find resources easier, and write them more efficiently. It didn’t write them for us. With the emergence of generative AI and Large Language Models, content literally writes itself. Even when students copied and pasted information from online, this was only possible to a certain extent, and couldn’t answer specific questions unless it had already been written online in that exact form.
Now, a student could ask ChatGPT to write an essay on an entire subject, without ever having to consider the question, or assimilate any information themselves.
Or, for that matter, actually getting the right information. LLM’s are built to be convincing, not to know what’s true and what isn’t. A professor of AI told the New York Times that if you don’t already know the answer to something, don’t trust AI to answer it – you won’t know what’s true and what isn’t.
Learning via researching and transcribing information, and in the process using your brain to understand that information whether you want to or not, may be a thing of the past.
If that’s the case, without changes to the education system school, even university, just got a lot easier. Learning, on the other hand, just got a lot harder.
Brave & Heart over and out.
Bonus
Skoda Is A Bear?
Of course, in the age of AI generated this and that, things are going to get a bit weird.
Someone decided it would be a fun and useful thing to do to ask AI to generate image of popular car brands as animals.
Skoda was portrayed as a “working class bear” – imagine a bear working in the post office in Zootopia – while Audi, for some reason, is a stag standing tall in a VERY tight red tracksuit.
Mini is pretty cool, it’s a little rabbit in a funky graffiti print outfit and white trainers.
To find out more on how you can retain your top talent, or how we can help you with digital solutions to your business and marketing challenges, check out our case studies.