GPT-4 Is Here, And Experts Say It Can Make You More Creative

The newly released GPT-4 artificial intelligence (AI) technology can do many things, but perhaps its most powerful ability is to help users be more creative. 

OpenAI recently announced the next-generation version of the artificial intelligence technology that runs its viral chatbot tool, ChatGPT.  It can produce more natural-sounding text and solve problems more accurately than its predecessor and can also process images in addition to text. Experts say the software is intended to aid human abilities rather than replace them. 

“GPT-4 ought to be both more useful and more creative than its predecessors in part simply because it is more intelligent or, at least, has a greater capacity for accuracy on cognitive tasks, whether or not one considers this sufficient for true intelligence,” Alex Kiefer, a research and development engineer at Verses.AI, a cognitive computing company, told Lifewire in an email interview. “More intelligence entails greater sensitivity to distinctions and a greater capacity to draw connections between disparate ideas or domains.”

Eerily Smart

GPT-4 has an uncanny ability to accomplish tasks that, until recently, would have seemed like science fiction. The new software can draft lawsuits and provide more precise responses to queries. One of the most impressive examples of GPT-4’s powers has come from an OpenAI video demonstration that showed how a drawing could be turned into a functional website within minutes. 

“GPT-4 is a large multimodal model, (accepting image and text inputs, emitting text outputs) that, while less capable than humans in many real-world scenarios, exhibits human-level performance on various professional and academic benchmarks,” OpenAI wrote on its website.

An AI engine like ChatGPT has access to a nearly immeasurable amount of information that it processes by sweeping the internet of data. It can combine all those facts in ways that make sense, Jamie Boote, a consultant at Synopsys Software Integrity Group, said via email. 

“By extending the capabilities of the engine, ChatGPT is better at synthesizing novel combinations of existing ideas into new combinations that make it more creative,” he added. “By refining the model and training sets to keep ChatGPT from spouting text that humans would interpret as garbage keeps it from crossing the line into nonsense more often.”

Creative humans thrive on new ideas and new combinations. ChatGPT can be an endless source of varieties to build on, Boote said. 

“The process from idea to fixed work is often a long and arduous one to the point that the copyright office doesn’t recognize ideas, only fixed works,” he added. “A generative AI can lift the burden between idea and work by filling in the blanks, fleshing out skeletons of outlines, and creating substance from what appears to be nothing.”

Powerful generative AI might boost human creativity by producing novel outputs that human beings on their own couldn’t, or likely wouldn’t have, produced, Kiefer said.

“In this way, humans might use AI to bootstrap new forms of creativity,” Kiefer added. “An example is GPT-4’s ability to summarize text using only words that start with a certain letter, or to quickly compose poems on very specific, including highly technical or scientific, subject matter – while it’s conceivable that a human being could learn to do these things, it’s probably much more likely that GPT will accomplish these, and potentially many other impressive feats, first.”

Still Limited

As impressive as GPT-4 is, the software still can’t do many things that humans are capable of. It can also produce false or dangerous answers. Even Sam Altman, CEO of OpenAI, said in a tweet that GPT-4 “is still flawed, still limited, and it seems more impressive on first use than it does after you spend more time with it.”

Boote said he’s not convinced that ChatGPT should be used unsupervised or create outputs that don't need to be vetted by experts. 

“Based on my experience with the engines, ChatGPT 4 still exhibits many of the same overarching flaws that ChatGPT 3.5 does, and this is an incremental step rather than a revolutionary one,” he added. “There will be a point where the internet will be dominated by AI agents like ChatGPT, but that’s still a future state rather than one we’ll have to worry about this week."