The Future of AGI: Rivalries, Risks, and the New Economic Reality

[Paul Ritzer]: It is just this completely wild unknown world we're heading into where basically these five companies are going to decide everything when it comes to the economy, business, and geopolitics.

Welcome to the Artificial Intelligence Show, the podcast that helps your business grow smarter by making AI approachable and actionable. My name is Paul Ritzer. I'm the founder and CEO of Smarter X and Marketing AI Institute, and I'm your host. Each week, I'm joined by my co-host and Smarter X Chief Content Officer, Mike Kaput, as we break down all the AI news that matters and give you insights and perspectives that you can use to advance your company and your career. Join us as we accelerate AI literacy for all.

Welcome to episode 207 of the Artificial Intelligence Show. I'm your host, Paul Ritzer, along with my co-host, Mike Kaput. We're recording on Monday, March 30th, 2026, right before 10 a.m. Eastern time. I don't know if we're getting new models this week, but there is a lot of chatter going on about what's coming up from all the labs. This episode, we're going to be setting the stage for what I think is going to be a pretty busy spring. In some ways, we might see some pretty rapid advancements from the models, and these labs are pushing out a lot of stuff. We're going to try and provide the context to what's going on and help people frame it into what it means for their careers and their businesses.

There's a lot happening. As we were getting ready for the show, even just two minutes before we came on, Mike and I were like, "Wait a second. Didn't this happen in 2024?" So, we're going to do our best to provide a little historical context into what's happening.

This episode is brought to us by AI Academy by Smarter X, which helps individuals and businesses accelerate their AI literacy and transformation through personalized learning journeys and an AI-powered learning platform. New educational content is added weekly so you always stay up-to-date with the latest AI trends and technologies. The AI for Departments collection features five course series and certificates designed to jumpstart AI understanding and adoption. We have AI for Marketing, AI for Sales, AI for Customer Success, AI for HR, and AI for Finance. Mike is wrapping up AI for Operations this week, so that one is coming soon.

If you join AI Academy or if you're already a member, those are all in there already. Tell your peers in your organization if they're trying to figure this out; there's a department series for them. These series are an ideal launchpad for organizations that want to level up their teams and accelerate AI adoption and impact. Mike teaches the AI for Sales series and is going to be sharing some insights towards the end of today's episode. Individual and business account plans are available now, or you can buy single courses and series for one-time fees. Visit academy.smarterx.ai to learn more.

Every week, we go through what we call our AI Pulse, where we take an informal poll of our listeners on how they feel about topics we talk about in that episode. Then we'll go through three main topics and then rapid-fire items.

AI Pulse Survey Results

[Paul Ritzer]: From episode 205—because we had an AI Answers episode for 206—the first question was: "OpenAI is building an enterprise deployment arm with private equity backing. What's your reaction?" This one looks like a perfectly split pie. 25% say it's a smart move because AI companies need distribution, not just models. 26% said they don't have an opinion. 28% said it was inevitable and every AI company will do this within a year. 20% said it's concerning because it blurs the line between AI vendor and consulting firm.

The second question was: "Anthropic's 81,000-person study found the number one fear is hallucinations, not job loss. Does that match your experience?" 43%, the largest percentage, said no, job displacement is still their top concern. 34% said yes, unreliability is the biggest barrier to trusting AI at work. 13% said neither, their biggest concern is something else entirely. 9% said they are not particularly worried about AI risks right now.

We also asked: "How many AI tools does your organization officially approve for employee use?" 45% said one to two tools. 34% said three to five. Only 15% said six or more. There was a small sliver that said none, AI is blocked or not addressed. If you're listening to this show and you work for a company that's blocking everything, there's a decent chance you might not be at that company very long. You might be looking for a new career opportunity where you get to apply everything you're learning in AI. You can participate in those each week at smarterx.ai/pulse.

The OpenAI vs. Anthropic Rivalry

[Paul Ritzer]: We're going to start off today with a topic that spun out of OpenAI cancelling their Sora app, the individual app. We zoomed out and said, "Let's talk about the bigger thing going on," because we touched on how OpenAI was refocusing their efforts. We're starting to get a more sense of why that's happening and where this is going. We wanted to frame it within the OpenAI versus Anthropic topic.

[Mike Kaput]: We had a major Wall Street Journal investigation this week that traces this OpenAI-Anthropic rivalry way back to basically almost a decade ago to a San Francisco group house in 2016 where multiple players were living. It reveals that this feud—and it very much is a feud—is shaping the future of AI and is as much about personal wounds and power struggles as it is about bigger picture topics of philosophy or safety.

This piece is based on interviews with current and former employees at both companies. Tensions started very early. After Anthropic CEO Dario Amodei joined OpenAI in 2016, he watched Elon Musk order layoffs in ways that he considered needlessly cruel. He also watched Greg Brockman float the idea of selling AGI early on to the nuclear powers on the UN Security Council. Dario, as early as 2016 or 2017, started considering that proposal tantamount to treason and nearly quit over it.

When Sam Altman took over OpenAI after Musk exited in 2018, things got more complicated. Altman made Dario a promise that Brockman and Ilya Sutskever would not be in charge, and then turned around and made conflicting promises to Ilya and Greg. As research into GPTs took off, Dario blocked Brockman from working on the language model project. Daniela Amodei, Dario's sister, offered to step down rather than let Brockman join. By 2020, relations had deteriorated to the point where Altman accused the Amodeis of plotting against him to the board.

This culminated in late 2020 when Dario, Daniela, and nearly a dozen employees left to found Anthropic. Before leaving, Dario wrote a memo arguing the ideal AI company would be 75% public good and 25% good for the market. Now, five years later, both these companies are valued at hundreds of billions of dollars and racing towards an IPO. In recent months, Amodei has escalated the conflict sharply. He compared the Altman and Musk legal battle to Hitler and Stalin fighting. He called Brockman's $25 million pro-Trump super PAC donation "straight up evil" and likened OpenAI to a tobacco company.

This is all happening as competitive pressures reshape both companies. OpenAI shut down its Sora video app, which was burning a million dollars per day and dropped to under half a million users. Fiji Simo, the head of applications at OpenAI, described Anthropic's gains in the enterprise market as a wakeup call and told staff, "The company cannot miss the moment because we are distracted by side projects." Paul, how much of this is just personal versus the bigger picture philosophy?

[Paul Ritzer]: It definitely seems like there's a lot of residual bad feelings. If you're relatively new to this, you've heard this story unfold, but these details are the most detailed unfolding of events I've seen. The reason we want to talk about it is because it's so relevant to everything else happening right now. You have this battle over government contracts where Anthropic was designated a supply chain risk. OpenAI steps in the day Anthropic is getting blackballed and says, "Hey, we'll take the contracts." For Dario, that is just daggers.

They're both racing to IPO this year. They're both being funded by a lot of the same people and companies. They're now in a battle for the enterprise. Every day I'm talking to leaders who are moving to Anthropic; it is a very common recurring theme.

OpenAI was created in November 2015 intentionally as a counterbalance to Google. It was Musk, Altman, Ilya, and Greg. They wanted to be the alternative to Google, which they considered the "evil empire." It quickly becomes not a nonprofit, which creates the friction between Musk and Altman that will go to trial soon.

A lot of this stems from Dario Amodei not getting the credit he thought he deserved. Greg Brockman was trying to get Dario and Daniela to join them. Greg and Daniela worked at Stripe together. Dario was at Google. Eventually, they come over. There is another name that matters: Holden Karnofsky. He was the founder of a philanthropy promoting effective altruism, the antithesis of techno-optimism. Karnofsky, who is Daniela's fiancé, is a major player. Brockman starts taking an interest in effective altruism.

In 2016, they were debating who should be told if they build AGI. Should they tell the American public, or talk to the government first? Dario argued it was better to go to the government first. By mid-2016, Dario joins the lab. This is when the layoffs happen, led by Musk. In fall 2017, Dario brings in an ethics adviser. This is when Brockman sees the fundraising idea to sell AGI to governments including China and Russia, and Dario calls it treason.

Musk exits in 2018. Altman takes over as CEO. They go down the for-profit path. Karnofsky, now married to Daniela, is on the OpenAI board. Tensions flare when researcher Alec Radford lays the groundwork for GPTs. Brockman wants a piece of this, and Dario, as research director, says no way. Daniela tells Brockman he cannot work on it and offers to step down as head of the project rather than allow him on it.

When Brockman says he and Altman are going to meet with Barack Obama, Dario gets cut out of the meeting. He's pissed. He eventually wants to report directly to the board or nothing. He thinks they need to go in the direction of public good. This leads to them leaving.

In 2024, Brockman took a leave. We eventually found out the sabbatical was a mutual agreement stemming from internal friction about his management style. September 2024 is when the o1 reasoning model comes out. Brockman takes his leave in August. Meera Murati, the CTO, leaves the week they announce the reasoning model, and then Greg comes back.

The friction exists because they all know each other. They all came up together. Each of these labs is working on dimensions of AI progress: agentic capabilities, computer use, continual learning, memory, reasoning, and recursive self-improvement. Recursive self-improvement is the idea that as labs automate AI researchers, those agents can work 24/7 and improve the models without human oversight.

There are five frontier labs. Tier one: Google DeepMind, OpenAI, and Anthropic. Tier two: Meta and xAI. Tier three: Microsoft, if they figure it out. Two of the tier-one labs are at war, and the tier-two xAI is suing OpenAI. These five companies are going to decide everything.

Politics also plays a role. DeepMind and Google try to stay neutral. OpenAI's president is the largest donor to the current administration. Anthropic is seen as the enemy of the administration right now. Meta and xAI are 100% in with the Trump administration. What happens if the power switches and the government doesn't award contracts to certain labs? The only ones left might be the politically neutral ones. Knowing these characters is extremely important because it's not a binary decision. There are layers to who you're investing in or whose technology you're using.

[Mike Kaput]: Whether you agree with the AI hype or not, all of these people have been taking the prospect of AGI seriously for ten years. Before they even had a business model, they were taking seriously who should have control of this technology.

[Paul Ritzer]: I listened to a Lex Fridman podcast with Peter Steinberger, who created OpenClaw. It sure sounds like he's selling to Meta. Zuckerberg was playing with OpenClaw and messaged him on WhatsApp. I can't see an alternative at this point unless Sam Altman pulls a rabbit out of a hat.

Details Leak on Anthropic’s New Model

[Mike Kaput]: Anthropic accidentally exposed details of an unreleased model nicknamed "Claude Mythos" through an unsecured content management system. Roughly 3,000 unpublished assets were accessible, including draft blog posts and documents about an invite-only CEO retreat. The leaked drafts describe Mythos as a new tier above Opus, with dramatically higher scores on coding, academic reasoning, and cybersecurity.

Anthropic confirmed the model is real and called it a step change. They warn that Mythos is far ahead in cyber capabilities and can exploit vulnerabilities in ways that outpace defenders. They planned to release it to cyber defense organizations first. They blamed the leak on human error in their CMS configuration.

At the same time, OpenAI says it has finished pre-training its next major model, code-named "Spud." Altman told staff he expects a very strong model within weeks that can "really accelerate the economy." Paul, when do you expect these to drop?

[Paul Ritzer]: If they've got stuff queued up in a CMS, it's ready. These things probably finished training months ago and have been in red teaming. You cannot make plans based on your current experience with these models. There is always a more powerful model in training. The labs have already seen 6 to 12 months ahead of what you know to be true.

I feel for the marketing team at Anthropic. I can't even fathom being the person that allowed that to happen. Part of this is a story about accidental disclosure, and part of it is how much easier discovery will be with agents running 24/7 looking for vulnerabilities.

Fortune informed Anthropic about the leak but has yet to publish the full information. There's probably a quid pro quo there. The bigger models worry me. Cybersecurity stocks slumped based on this news. CrowdStrike, Palo Alto Networks, and Zscaler dropped about 6%. It's funny because we've predicted this for two years, but that's how Wall Street works.

Regarding "Spud," Altman said the company would be renaming the product organization to "AGI Deployment." They truly think we are approaching AGI. In July 2024, OpenAI shared five levels of progress:

  1. Chatbots (conversational language)
  2. Reasoners (human-level problem solving)
  3. Agents (systems that take actions)
  4. Innovators (AI that aids in invention)
  5. Organizations (AI that can do the work of an organization)

In summer 2024, they thought they were at level one. By September 2024, they reached level two with the o1 model. We are now in the midst of takeoff with agents (level three). I think we will be at level four by this fall. We went from level one to level four in basically 20 months.

Brutally Honest CEO Perspectives

[Mike Kaput]: Uber CEO Dara Khosrowshahi broke an unwritten rule in tech. He said he has personally heard executives privately admit the scale of AI disruption and then watch those same people go on TV and say everything will work out fine. He said being honest scares investors, but he estimates AI will replace the work 70 to 80% of humans do within the decade. When asked what Uber's 9.5 million drivers do next, he said, "I don't know."

At the same time, PWC's US CEO Paul Griggs told the Financial Times that partners who are "not paranoid about being AI-first" will be replaced. He said an employee who thinks they can opt out of AI is "not going to be here that long." PWC cut 5,600 staff last year and is shifting services into AI-powered subscription tools. Paul, is the dam starting to break?

[Paul Ritzer]: What executives are saying privately and what they're saying publicly have been two different things for a long time. It's getting hard not to say it out loud on earnings calls. I think AI-forward managers and above who have domain expertise and AI literacy are going to be in good shape. Companies will look for that talent.

However, professionals who are resistant to learning AI are going to have a very difficult time remaining employed. It is a brutal reality. My biggest concern is entry-level work. I don't know what you're going to hire those people for when the layer above can do all the tactical work by prompting a system.

There was a National Bureau of Economic Research paper from 2026 that examined AI productivity. It found that while adoption is widespread, larger companies are focusing on labor cost reduction. It suggests AI currently functions more as a tool for task enhancement, but a significant reallocation of labor is underway. They interviewed CFOs, who are generally not the people with the greatest comprehension of AI capabilities. They might not even be aware of the reasoning or agentic advancements. You have to piece together your own story.

[Mike Kaput]: I appreciate the honesty from Paul Griggs. I know of organizations where people are already complaining about employees who aren't embracing this stuff. It's better to express that this is a condition of employment.

[Paul Ritzer]: I would much rather they just said it than pretend like it's not going to happen. Pretending it's going to be okay because "it always has been before" is either choosing to lie or being ignorant to how different this transformation is.

Rapid Fire: Politics, Security, and Apple

[Mike Kaput]: Federal judge Rita Lynn issued a preliminary injunction blocking the government's supply chain risk designation against Anthropic. She wrote that nothing supports the "Orwellian notion" that an American company may be branded a potential adversary for expressing disagreement with the government. This blocks 17 federal agencies from enforcing the ban. The Pentagon's Emil Michael called this a disgrace and said they are not backing down.

[Paul Ritzer]: It seemed like this was just a vendetta. It's not really about the technology. It's becoming increasingly political. I'm sure this will take forever to play out, but in the meantime, the government is going to keep using their tech anyway.

[Mike Kaput]: Senator Bernie Sanders and Representative Alexandria Ocasio-Cortez introduced the AI Data Center Moratorium Act, which would pause new data center construction until federal AI legislation is passed. Meanwhile, President Trump has appointed Mark Zuckerberg, Jensen Huang, Marc Andreessen, and Sergey Brin to a new council of advisers on AI. Notable absences include Sam Altman and Elon Musk. David Sacks is co-chairing this council.

[Paul Ritzer]: A pause is not going to happen. The bill is to raise awareness, not to get legislation passed. I also wouldn't hold my breath on federal legislation. This council has some big names, but we know almost nothing about it. There's also a new pro-AI PAC called Innovation Council Action, led by a former Trump staffer, with a plan to spend more than $100 million. Between the different groups, there's almost $300 million in ads coming about AI deregulation.

[Mike Kaput]: Andrej Karpathy flagged a "software horror" where malicious code was slipped into an open-source package called LightLLM, which has 97 million downloads per month. It was live for less than an hour, but it stole passwords and API keys. AI agents are going to make risks like these much worse as they start installing software and managing systems on their own.

[Paul Ritzer]: I have no idea how to be sure you're not downloading something poisoned. Most people have no idea. Cybersecurity is a safe profession to go into for the next decade. The surface areas where you can be attacked are endless. Enterprise and human friction might be the only thing that saves us here because it's going to take a while to figure this all out.

[Mike Kaput]: Apple is planning to open up Siri to rival AI assistants in iOS 27, ending ChatGPT's exclusive role. Users will be able to route Siri queries to Google Gemini or Anthropic Claude. Apple is also building a standalone Siri app to transform it into a system-wide AI agent.

[Paul Ritzer]: Apple is the wild card. They have trust and access to all your data. It seems like a smart strategy to let everyone else spend billions building models while they serve them up to billions of people. Does this make Perplexity irrelevant? If I can just choose whatever model I want through my iPhone, why would I need Perplexity?

Smarter X Use Case Spotlight

[Mike Kaput]: We want to give a quick look under the hood at real AI use cases we are exploring at Smarter X. Paul, you've been working on AI learning journeys.

[Paul Ritzer]: I spend a lot of time on the vision side. I tell the team we're not in the business of selling courses; we're trying to power transformation. You need assessments, executive briefings, and communications plans. I've been devising an "AI Transformation System." I had sketches I couldn't visualize, so I wrote a project brief for a designer—except the designer was AI.

The prompt was 1,000 words. Gemini gave me a useless infographic. Claude gave me a solid V1 with drag-and-drop capabilities. GPT-4o gave me a really solid interactive prototype. I had an interactive system in seven minutes that would have cost tens of thousands of dollars to build with a developer. Take the time to write a prompt as though you were giving it to an internal person.

[Mike Kaput]: I've been working on AI-powered slide creation. I have to create hundreds of slides for our AI Academy courses, and it's not intellectually rewarding work. I finally got Claude Code to do this with a high degree of fidelity. I provided an excellent set of example files and put it into "planning mode" first. Now, Claude Code can take my scripts, put them into presenter notes, and build the slides with placeholders. It took 20 minutes instead of hours.

[Paul Ritzer]: It's interesting how different our workflows are. Mike is a script guy; I build slides first to form my thoughts. But it shows that personalized use cases are critical. When you drill in and create specific workflows, you unlock hundreds of hours of productivity.

AI Academy Spotlight: AI for Sales

[Mike Kaput]: This week we are spotlighting our AI for Sales certificate series. Here are three big takeaways:

  1. Sales reps only spend about 30% of their time actually selling. Use the "checklist test": if you can write the steps out for a task, it should be automated.
  2. Audit your existing tech stack before buying anything new. Look to Salesforce, HubSpot, or Microsoft first.
  3. Stop prompting like a search engine. Use structured prompts with roles, tasks, context, and examples.

AI Product and Funding Updates

[Mike Kaput]: To wrap up:

[Paul Ritzer]: Predicting how human brains respond to sights and sounds? I can't come up with a positive use for that.

[Mike Kaput]: One final reminder: take the AI Pulse survey at smarterx.ai/pulse. Paul, thanks for breaking down a busy week.

[Paul Ritzer]: Next Tuesday, we'll have a Q1 AI Trends Briefing for business. Keep an eye out for that. Have a great week, and stay curious.

Key Takeaways