Talking Heads to Thinking Machines
A sharp turn in AI capability may be near. Here's what's coming around the bend.
Our Futures team at Weber Shandwick has run dozens of hands-on AI labs for clients this year. One of the most engaging tools in portfolio reviews, the LLM that creates the most energized reaction in the room, is Opinionate.io.
This genAI-powered bot debates any prompted topic, from the obscure ("Is cereal soup?") to the serious ("Should I take Humira or Enbrel for my rheumatoid arthritis?"). Two models give opposing perspectives. A third one, the moderator, selects the winner. It happens like magic, an exchange happening right before your eyes.
Opinionate feels like a nascent prototype, a glimpse into the future, like Blogger for web publishing in 1999. Its stark, functional design emphasizes utility over aesthetics, making it feel more like a work-in-progress than a reliable, finished product.
Its appeal rests on showing the potential of AIs that reason. These new intellectual powers will soon become more accessible.
![](https://substackcdn.com/image/fetch/w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6a5304fc-eb35-4336-bc8d-e8281b442367_1402x726.png)
Industrial-Strength Reasoning Machines
Word recently leaked that OpenAI is developing an industrial-grade reasoning engine codenamed Strawberry. Documents reviewed by Reuters describe models designed to autonomously and reliably navigate the internet to find answers.
For a platform that is 18 months old, ChatGPT has shown remarkable reasoning abilities, solving complex tasks as effectively as a highly educated human. It has achieved scores in the 90th percentile on the SATs, mastered Sommelier exams, and even passed the Turing Test.
OpenAI has far more ambitious plans in the making. They envision powering autonomous agents capable of performing tasks over extended periods of time (unlike the immediate replies seen now). They will autonomously invent new technologies and operate as organizations that mirror their human counterparts. Sam Altman says:
"For me, AGI is basically the equivalent of a median human that you could hire as a co-worker. And then they could say do anything that you'd be happy with a remote coworker doing behind a computer, which includes, learning how to go be a doctor, learning how to go be a very competent coder.
There's a lot of stuff that they’re capable of getting good at. AGI is not any particular milestone but a meta skill of learning to figure things out and then it can go decide to get good at whatever you need."
Altman doesn't explicitly state here that autonomous AI teams will solve problems defined by us. This is when things get truly interesting — and, as they say — disruptive.
To add, Eric Schmidt, former CEO of Google and co-author of The Age of AI: And Our Human Future, discusses the depths of superintelligence coming.
Paraphrasing the video below, he said:
“I think the thing that to understand is that we’re going to have a very different world and it’s going to happen very quickly for the following reason.
The systems will get so good that you and I everyone in this audience and everyone in the world will have access to a essentially a polymath, as in the historic polymath of old.
So imagine if you had Aristotle to consult with you on logic and your Oppenheimer to consult with you on physics, not the person but rather than knowledge and that kind of scaling intelligence.
These sort of truly brilliant people who were historically incredibly rare. They’re their equivalents will become generally available.”
Just More AI Hype?
Skeptics note that amidst all the hype and projection of promise, enterprise-grade generative AIs still aren’t trustworthy, and it’s far too early to shout about agents. There are bumps that stunt progress —maybe walls—like data deficits, hallucinations, computational limitations, and ecological impact.
Leopold Aschenbrenner makes a compelling case that blasts through perceived barriers. In the widely discussed paper "Situational Awareness," Aschenbrenner shows engineers generating an orders-of-magnitude (OOMs) escalation of computing power, indicating a ~100,000x effective compute scale-up in progress.
This leap suggests another GPT-2-to-GPT-4-sized qualitative jump over four years. Following OOM power curves, he projects that by 2025 or '26, AIs will outpace college graduates and exceed the intelligence of PhDs by 2030.
Whether you accept Aschenbrenner's thesis or not (as many do), his comments are also worth mulling:
“Everyone is now talking about AI, but few have the faintest glimmer of what is about to hit them. Before long, the world will wake up.
“But right now, there are perhaps a few hundred people, most of them in San Francisco and the AI labs, that have situational awareness. Whether these people are also right about the next few years remains to be seen. But these are very smart people—the smartest people I have ever met—and they are the ones building this technology.
Perhaps they will be an odd footnote in history, or perhaps they will go down in history like Szilard and Oppenheimer and Teller.
If they are seeing the future even close to correctly, we are in for a wild ride.”
AI's power surge has big implications for leadership, finance, strategy, and HR in established organizations. For entrepreneurs, the curve gives us a glimpse into Altman's seemingly absurd statement that there will be one-person, billion-dollar companies.
Behind the hype and timing of advances are cultural, legal, and ethical considerations that all leaders will inevitably confront. Automation advances will force us to rethink what it means to be human psychologically and spiritually.
How to Think About AI’s Progression
With so much happening simultaneously, how do we begin to navigate changes AI titans forecast? For starters, frameworks can help us understand different AIs and how they work together, allowing us to reflect on and anticipate their effects. Here’s a basic way of thinking about it.
Artificial intelligence, not a newcomer to the scene, has shaped our lives for decades. From the expert systems of the 1980s to search engines of the 1990s and social networks today, we've felt both the promise and pitfalls of algorithmic living. Leaps in computing power, data, and algorithms have ushered in an emerging AI mix wielding different intelligences.
Recommenders are digital librarians whispering in our ears through social media feeds and e-commerce suggestions. They curate our world, filtering vast information into personalized streams. Yet, as they guide us toward the familiar, they risk trapping us in echo chambers, narrowing our horizons, and eroding our capacity for critical thought.
Then came the Talking Heads, the eloquent AI parrots that predicted and mimicked human speech with startling accuracy. Large language models, like ChatGPT, generate anything from code to poetry. Will they be our partners, amplifying our creative pursuits, or will they simply replace us, leaving us speechless and obsolete? We don't know yet, but a dislocation of work and knowledge creation hangs in the balance.
Finally, on the horizon loom the Thinking Machines, the autonomous agents, the problem solvers, the decision-makers. These are the true wildcards, promising to cater to our needs and act on our behalf. They are the stuff of utopian dreams and dystopian nightmares, the potential saviors or destroyers of us.
Optimists envision a world where these AI agents tackle our most pressing challenges —from climate change to disease—and free us to pursue higher callings. Thinking machines could catalyze a new renaissance and unprecedented human flourishing.
Yet, skeptics warn of a darker future, where our agency and individuality are sacrificed at the altar of automation. We risk becoming mere cogs in a machine, our lives dictated by lines of code. The question remains: will these thinking machines elevate us or permanently displace us?
We face a highly uncertain phenomenon. The next five years will refine our view and, as some have warned, change society's trajectory for decades.
A Crush to Comfort Zones
The thesis of my book suggests courageously facing a disruption of this magnitude starts in our minds. It requires widening your lens. Viewing this moment with clear eyes can be a springboard for growth and meaningful contribution.
A normalcy bias stands in the way—the deeply ingrained view that things will continue as they always have, despite knowledge of breaks in precedent. In my client work, I see it manifest as busyness (I can't focus on AI when I have a business or projects to run), skepticism (AI is useless or a bubble), and apathy (I hope this just goes away).
A Bain report suggests normalcy weighs on progress. Unlike traditional surveys, Benedict Evans noted that Bain split generative AI pilots, experiments, and trials from full deployment. Their findings suggest that while there is great interest and many tests in motion, few trust generative AIs for business change just yet.
I see it in a more nuanced, human way. We've found a conflict of vision between those running companies and those deep inside them desperate to break new ground. A stark divide becomes clear: Those holding the line (maybe for good reason) and those energized by the opportunity to shape their future.
Either way, the economic and technical forces behind AI will be comfort crushers for all of us. Social media proved that you don't have to care about a technological shift to be affected by it. Understanding the potential impact of AI is not optional but a necessity.
We face a future where AI isn't just a tool but an active participant in our world. Will we wait and see (or look away) or grab an opportunity with both hands to lead through better intelligence?
There are many problems to sort out and, maybe soon, extraordinary capabilities to solve them. Getting caught in a normalcy trap is a no-man land. A different future is being scripted now.
What comes from it will be a wild ride. Time will tell if it's a thrill ride or a house of horrors we can't manage.
In my previous Substack post, I previewed Perspective Agents becoming a “learning engine,” with the book content summarized in a prompt guide to be explored through a tailored GPT in OpenAI’s store.
Check it for more on the progression of agents and how they influence politics, culture, and business. If you like it, spread the word. Thanks for reading.