OpenAI’s “Intelligence Age” Is Already Here, It’s Just Not Evenly Distributed

OpenAI’s “Intelligence Age” Is Already Here, It’s Just Not Evenly Distributed

OpenAI has been busy lately. They're not just launching new models and features; CEO Sam Altman also shared his vision of the future.

Sam Altman's perspective on AI

This week he shared a blog post on the "Intelligence Age." It's not sci-fi; it's a roadmap for AI and humans to solve global challenges. Altman sees AI as a partner amplifying human capabilities, potentially revolutionizing industries, and economies. But he's not ignoring the challenges. Altman acknowledges the massive changes AI will bring, including job disruptions.

Path to Artificial General Intelligence

OpenAI's ultimate goal is to create artificial general intelligence (AGI), an AI that can outperform humans on most tasks. According to Bloomberg, in July they shared internally their definitions for five levels of AGI, resembling similar schemes for understanding progress toward fully autonomous vehicles. Many believe we have just reached level 2 - Reasoners with o1-preview, an early version of the upcoming full o1 model.

• Level 1 - Chatbots: AI with conversational language

• Level 2 - Reasoners: human-level problem-solving

• Level 3 - Agents: systems that can take actions

• Level 4 - Innovators: AI that can aid in invention

• Level 5 - Organizations: AI that can do the work of an organization

Fun fact: they don't refer to the possibility of AI systems becoming autonomous - able to act on their own initiative without human direction.

Let's look at two recent announcements that bring this vision to life:

o1-preview (aka Strawberry): The AI model that can "think and reason"

Meet o1-preview, OpenAI's new thinking and reasoning model. While GPT-4o is still a fantastic tool for everyday needs due to its speed and versatility, o1 excels in tasks requiring deep reasoning, planning, and multi-step solutions. o1 tackles complex problems in science, math, and coding with incredible skill. It solved 83% of International Mathematics Olympiad qualifier problems, a leap from its predecessor's 13%. Now, imagine a future with a PhD-level assistant that can analyze data, predict trends, and develop strategies.

But there’s a caveat: this model takes longer to think through the prompts. This extra effort helps break down complex tasks, refine their approach, and catch their own mistakes. Also, it can’t access the web and doesn’t offer file uploads (yet).

Here are quick tips to prompt o1-preview:

• Keep it simple and direct

• Don’t use chain of thought prompts (it’s already built in)

• Use delimiters like quotation marks, tags or section titles

• Don’t overload it with information

Advanced Voice Mode: AI gets personal

OpenAI also announced the rollout of Advanced Voice Mode for ChatGPT that sound strangely like a human (yes, it says “hum”, laughs and changes the tone when talking about a difficult topic). It can store custom instructions, memory of the behaviors the user want to exhibit, and has nine unique voices with their own personalities. This isn’t just fun; it’s about creating more natural, engaging AI experiences. For marketers, it opens possibilities in personalized customer interactions but also challenges ensuring our brands are well represented via audio.

The bigger picture

As we enter the Intelligence Age, the potential of AI to transform marketing is huge. From hyper-personalized campaigns to AI-assisted strategies, our toolkit is evolving rapidly.

As Professor Ethan Mollick said, “as these systems level up and inch towards true autonomous agents, we’re going to need to figure out how to stay in the loop - both to catch errors and to keep our fingers on the pulse of the problems we’re trying to crack.“ Onward.