With each passing day conspiracy theorists have more stuff to spook the normies.
I bet it was fun reading about the digital collapse of the year 2000 and robots taking over the world in the future. But have you seen DALL-E-2?
Now we have another algorithm making the news — ChatGPT. Let’s talk about it.
What is ChatGPT?
Let’s start off easy by figuring the thing out.
ChatGPT is basically a chatbot powered by a strong AI algorithm that can generate — well, answers.
Ask it a question, give it a premise for a story, command it to write some code, and the technology will provide the somewhat consistent results. That’s the short answer.
For more context: it was launched on November 30th as a part of a technology set developed by San-Francisco-based startup OpenAI. Some time ago, the company received an investment of $1 billion from Microsoft and now headquarters at their Pioneer building.
And now Microsoft is planning to put in $10 billion more. That poses the question — what is so special about ChatGPT?
Feeding off a vast database of books, online essays, articles, and other sorts of writing, ChatGPT can converse and produce readable texts on any topic.
The name hints at the method of manipulating the language model — you communicate with the algorithm and it responds with the desired content. You can then change those results by asking additional questions or providing notes.
Finally, I can feel like an editor for once!
What differentiates ChatGPT AI from the language models released before it is the sheer number of parameters the tool works with. As OpenAI claims, GPT-3 has 175 billion parameters to produce the content. Before that, most of the modules never exceeded a billion parameters.
That’s what collaboration with Microsoft can do. Supporting that technological potential requires a lot of computing power, which the company provides in the form of Azure infrastructure with over 285 thousand CPU cores and 10 thousand GPUs.
So what’s the fuss about?
It is surely not the first piece of technology that creates texts, right?
Why is everybody so hyped about it? Because it simply works so well.
At first, people started playing around with it, trying to see if they could trick it into writing nonsense and making mistakes. Some create poems, songs, or scripts.
But then things got more serious.
On one occasion, a user has asked the algorithm to write his cover letters for job applications. And get this — the letters were good enough so that the person actually got responses from recruiters!
Then, the students started using it to write their papers. Though the urge to cut corners is understandable for anyone who has ever been tasked with writing dozens of pages each semester — come on guys.
You’re putting term paper writers out of work.
It has gotten to the point where you don’t even need to know a skill to produce something — some users were able to make the tool code without knowing a single thing about it. In one instance, a person with no programming experience whatsoever was able to get a Python script which could encrypt users’ data.
And the coolest use case of ChatGPT in media, at least for me — video-game dialogue generation. When it comes to large RPG games that you can replay countless times, the dialogue in those games becomes stale pretty fast. Since it is all pre-recorded, you end up hearing it a lot of times, reaching the point where you just ignore or skip right through the lines.
Now, ChatGPT automation can change that by allowing characters to generate unique responses to your questions on the fly, while still providing valuable information. That adds on to the immersiveness of your adventures by a lot, don’t you agree?
Those are just a couple of examples, but they are a strong proof of ChatGPT potential.
Is ChatGPT a cure for all?
Maybe you got the idea that it is an ultimate tool for all things speech and text. It is a definite step up in quality from the things I had an opportunity to try out during my college years, it is still unclear how far it can go beyond its current potential.
Those trembling with English degrees may not worry just yet: though it manages to fool plagiarism checkers, it still can’t generate correct and compelling abstracts, as claimed by Arvind Narayanan, a computer science student at Princeton University.
That drawback is the reason ChatGPT automation capabilities are not ready for the world of commercial writing as well. Dealing with editors and clients' notes is hard enough for humans — now imagine a machine trying to solve that riddle. The automated approach takes away the sophisticated control over the result and understanding between the writer and its client.
Tell us what you think on AI automation!
Some also claim that ChatGPT can prove to be an effective search tool. Just like Google, the tool can source information and generate results referencing that information with an amazing efficiency, so why not?
But if we look at the search engine trends from the last couple of years, we can see the preference over trusted sources with reliable information dominating other technological tweaks. In other words, upon updating Google’s page ranking, the developers try to look for genuine content from competent and reliable sources.
And ChatGPT can do none of that. Simply saying, it can write a great article about the flatness of our mother Earth. And that’s the issue.
You can get a deeper dive into the reasoning with this thread:
In order to understand why ChatGPT can't replace Google Search, it's useful to understand the early days of web search and the role that PageRank played. 1/n— MMitchell (@mmitchell_ai) December 20, 2022
But let’s not dwell on here any longer. Is there any business value of ChatGPT for media companies?
How ChatGPT helps with media automation
I wouldn’t be the first to point out the gripes of media industry effectiveness. While you can optimize data and resource management pretty easily with CRM tools and stuff like that, streamlining other aspects of your process can be more challenging.
Take content production, for example. There’s almost no getting around the time it takes to produce the majority of the clips, movies, and videos, and podcasts, but there is a way to take some of the less creatively intensive workload off your shoulders and therefore cut the production time.
Hear me out.
Producing something like a long-form interview requires a great deal of effort from human creatives. Booking a guest, researching the topic, scripting the whole thing, then shooting and editing the footage.
But there are also things that are more tedious than creative: for example, you want to have an intro to your video interview featuring the best moments.
Or you want that compilation put on social media to promote your content. That clip would need to be cropped to fit the portrait aspect ratio, which is the most prevalent on popular platforms like TikTok or Instagram.
It’s all about the time you need to spend on those tasks, right? That’s a great use case for AI media automation software, and such tools as ChatGPT.
You see, you can apply something of that sort into your production workflow, so that it could pick out the quotes from the footage. Then, you can have the software edit the clips with the quotes into a coherent compilation.
The automation of said process helps you increase the workflow effectiveness by a large margin: in some cases, the automated content production is 50x faster than the human editor.
If you're looking to try out automating some of yout production workflows, feel free to reach! Email us at email@example.com or use the form below to ask questions, share your ideas, or get a free trial!