Chat robot typing on a laptop sitting on a table.
PHOTO: Hamid

The Gist

  • Welcome, GPT-4. OpenAI has released GPT-4, which it says exhibits "human-level performance on various professional and academic benchmarks."
  • Brand alignment? GPT-4 has improved "steerability" which allows developers to tune the model's tone and behavior with specific directions, making it easier for marketers and CX designers to control a custom chatbot's personality in line with brand voice, style and policies.
  • Walking the GPT walk. OpenAI has been using GPT-4 internally with great impact on functions such as support, sales, content moderation, and programming, and is using it to assist humans in evaluating AI outputs.

Marketers and customer experience professionals: brace yourself. GPT-4 is here. And it's better than GPT-3, the creators say.

OpenAI, which created the wildly-popular generative artificial intelligence chatbot, ChatGPT, released today GPT-4, which it calls the latest milestone in OpenAI’s effort in scaling up deep learning.

Wondering how you can use this in your marketing or customer experience worlds eventually? Just let OpenAI tell you: "We’ve also been using GPT-4 internally, with great impact on functions like support, sales, content moderation, and programming. We also are using it to assist humans in evaluating AI outputs, starting the second phase in our alignment strategy.”

GPT-4 is a large multimodal model that accepts image and text inputs, emitting text outputs. OpenAI officials say it is less capable than humans in many real-world scenarios, but also exhibits "human-level performance on various professional and academic benchmarks."

OpenAI is releasing GPT-4’s text input capability via ChatGPT and the API (with a waitlist). The company is preparing the image input capability for wider availability and is collaborating closely with a single partner to start. It is also open-sourcing OpenAI Evals, its framework for automated evaluation of AI model performance, to allow anyone to report shortcomings in our models to help guide further improvements.

UPDATE: OpenAI Founder Showcases GPT-4 Capabilities

Greg Brockman, president and co-founder of OpenAI, conducted a developer demonstration in a YouTube video streamed late afternoon ET March 14.

"Honestly, it's kind of hard for me to believe that this day here," Brockman said. "OpenAI has been building this technology, really since we started the company but for the past two years we've been really focused on delivering GPT-4. That started with rebuilding our entire training stack, actually training the model, and then seeing what it was capable of, trying to figure out its capabilities, its risks, working with partners in order to test it in real-world scenarios, really tuning its behavior, optimizing the model, getting it available, so that you can use it."

Here's the video:

Related Article: ChatGPT: What You Need to Know

GPT-4 Wins on Content Summary Prompts

Brockman showed in his platform's "Playground" during his video GPT-3.5's tendency to "give up" on tasks like, "summarize this article into a sentence where every word begins with G." It came back with a sentence but not all with Gs. "For the most part, GPT-3.5 just gives up," Brockman said.

With the exact prompt in GPT-4, the model came through:

GPT-4 response: "GPT-4 generates groundbreaking, grandiose gains, greatly galvanizing generalized AI goals."

Brockman response: "AI doesn't count! That's cheating!"

GPT-4 response: "Gigantic GPT-4 garners groundbreaking growth, greatly galvanizing global goals."

Brockman response: Let's do the same thing with A. ("GPT-4" is fine.)

GPT-4 response: "GPT-4 achieves advanced advancements, amplifying artificial aptitude astoundingly."

Fun as this demo was, marketers and customer experience professionals will likely start to imagine this into their own workflows and creative strategies. What comes to mind here? Brainstorming subject lines. Customer success surveys question brainstorms. Some customer data analysis, perhaps. 

Related Article: ChatGPT Is All the Rage but Don't Stop Learning Just Yet

ChatGPT and Brand Voice, Style Up Next

Looking for gains in brand voice and style in this new artificial intelligence model released today by OpenAI? It may be soon arriving with GPT-4. 

Andrew Frank, VP, distinguished analyst at Gartner, noted in an interview with CMSWire on March 14 the improvement in what OpenAI calls “steerability,” which refers to a new API capability called “system” messages that allow developers to tune the model’s tone and behavior with specific directions.

"This will make it easier for marketers and CX designers to control a custom chatbot’s personality in line with brand voice, style and policies," Frank said. "OpenAI says these capabilities are coming to ChatGPT soon."

OpenAI's wasn't the only news in AI circles the past few days.

Microsoft has described its formula for integrating an organization’s own data into a ChatGPT-based chatbot using Azure Cognitive Search, paving the way for more personalized service and support, according to Frank.

"Not to be outdone so easily," he added, "Google is promoting its generative AI’s integration into gmail and docs in Workspace, assuring that competition will keep the generative AI race alive."