5 Steps That OpenAI Thinks Will Lead to Artificial Intelligence Running a Company

Earlier this month, Bloomberg reported that OpenAI had defined five distinct stages of innovation in AI, from rudimentary chatbots to advanced systems capable of doing the work of an entire organization. These stages could inform OpenAI's future plans as it works toward its ultimate goal of building artificial general intelligence, an AI smart and capable enough to perform all of the same work as a human. According to the Bloomberg report, OpenAI's leaders shared the following five stages internally to employees in early July during an all-hands meeting: Stage 1: "Chatbots, AI with conversational language" Stage 2: "Reasoners, human-level problem solving" Stage 3: "Agents, systems that can take actions" Stage 4: "Innovators, AI that can aid in invention" Stage 5: "Organizations, AI that can do the work of an organization" On July 23, the company posted briefly about the topic on X: "We are developing levels to help us and stakeholders categorize and track AI progress. This is a work in progress and we'll share more soon." Olivier Toubia, the Glaubinger Professor of Business at Columbia Business School, believes the five steps more closely resemble a plan to make human workers obsolete than a roadmap to artificial general intelligence. With the exception of reasoning, he says, all of the outlined stages are more focused on business uses than they are on the actual science. Toubia broke down what entrepreneurs need to know about OpenAI's five stages: Stage 1: Chatbots Bloomberg reported that OpenAI told employees that the company is still currently on the first stage, dubbed Chatbots. This stage is best exemplified by OpenAI's own ChatGPT, which shocked the world with its ability to converse in natural language when it was released in late 2022. Many organizations are using chatbots to enhance their internal productivity, Toubia says, while others are using the tech to power outward-facing customer service bots. While these chatbots may seem superhumanly smart at first glance, they're smoother talkers than they are operators. Chatbots will often make up and present false information with full confidence, and unless they've been set up to retrieve info from a businesses' data center, they don't have much commercial utility. Even Sam Altman has referred to the current iteration of ChatGPT as "incredibly dumb." Stage 2: Reasoners OpenAI told employees that it is close to creating AI models that could be classified in its second stage: Reasoners. According to Bloomberg, Reasoners are systems "that can do basic problem-solving tasks as well as a human with a doctorate-level education who doesn't have access to any tools." Last week, Reuters reported that OpenAI is currently at work on a new "reasoning" AI model, code-named Strawberry, focused on capabilities like being able to plan ahead and work through difficult problems with multiple steps. Reuters reported that leaders in the AI space believe that by improving reasoning, their models will be empowered to handle a wide variety of tasks, "from making major scientific discoveries to planning and building new software applications." Stage 3: Agents OpenAI doesn't believe that innovation in artificial intelligence has reached the Agent stage, which it refers to as "systems that can take actions on a user's behalf." Outside of OpenAI, much has been made of the potential value of digital workers who can operate autonomously, but few companies have wholeheartedly embraced the concept of AI Agents. Lattice, a popular HR software provider, recently announced plans to onboard AI Agents directly into a company's org chart, but scuttled the idea after online backlash. "From what I understand," says Toubia, AI Agents "could replace you for a few days when you go on vacation." Such an Agent would act as a proxy for vacationing employees, picking up the slack, completing simple tasks, and keeping vacationers updated on what happened while they were away. "I have a bit of a cynical view on this one," Toubia says, "people will welcome an Agent that's going to let them go on vacation more often, but based on the next steps, the goal is not just to replace you when you go on vacation, it's to replace you altogether." Stage 4: Innovators According to Bloomberg, Innovators refers to "AI that can aid in invention." In some ways, Toubia says, AI Innovators are already here. They're helping people generate ideas, write code, and create art. "With a bit of guidance," he says, "you can get ChatGPT to come up with ideas for a new app or a new digital product, and then create code and promotional materials." Because of this, Toubia predicts that Innovators, as defined by OpenAI, will mostly come in the form of AI systems specifically developed to help prototype, build, and manufacture physical products. Stage 5: Organizations In OpenAI's proposed final stage of artificial intelligence innovation, AI systems will become advanced and smart enough to do the work of an entire organization. Toubia says this should be a wake-up call for managers, who may have previously considered themselves safe from being replaced by AI, adding that even a company's founders could be considered expendable if the system finds that they're standing in the way of true efficiency. Toubia worries that by classifying Organizations as the final step in its "roadmap to intelligence," OpenAI may be tipping its hand regarding its ambitions. "This really seems to be a roadmap toward taking over the world," he says, "replacing complete organizations and making humans obsolete in the process." Going forward, he says, it may be CEOs who need to justify their paychecks.

AI in Business: Maximizing Gains and Minimizing Risks

Over half of CEOs recently surveyed by Fortune and Deloitte said that they have already implemented generative artificial intelligence in their business to increase efficiency. And many are now looking to generative AI to help them find new insights, reduce operational costs, and accelerate innovation. There can be a lot of relatively quick wins with AI when it comes to efficiency and automation. However, as you seek to embed AI more deeply within your operations, it becomes even more important to understand the downside risk. In part, because security has always been an afterthought. Security as an afterthought In the early days of technology innovation, as business moved from standalone personal computers to sharing files to enterprise networks and the internet, threat actors moved from viruses to worms to spyware and rootkits to take advantage of new attack vectors. The industrialization of hacking accelerated the trajectory by making it possible to exploit information technology infrastructure and connectivity using automation and evasion techniques. Further, it launched a criminal economy that flourishes today. In each of these phases, security technologies and best practices emerged to address new types of threats. Organizations added new layers of defense, often only after some inevitable and painful fallout. More recently, internet of things devices and operational technology environments are expanding the attack surface as they become connected to IT systems, out to the cloud, and even to mobile phones. For example, water systems, medical devices, smart light bulbs, and connected cars are under attack. What's more, the "computing as you are" movement, which is now the norm, has further fueled this hyperconnectivity trend. Organizations are still trying to understand their exposure to risk and how to build resilience as pathways for attackers continue to multiply and create opportunities for compromise. Risk versus reward The use of AI adds another layer of complexity to defending your enterprise. Threat actors are using AI capabilities to prompt users to get them to circumvent security configurations and best practices. The result is fraud, credential abuse, and data breaches. On the flip side, AI adoption within enterprises also brings its own inherent and potentially significant risks. Users can unintentionally leak sensitive information as they use AI tools to help get their jobs done. For instance, uploading proprietary code to an AI-enabled tool to help identify bugs and fixes, or company confidential information for assistance summarizing meeting notes. The root of the problem is that AI is a "black box," meaning there's a lack of visibility into how it works, how it was trained, and what you are going to get out of it and why. The black box problem is so challenging that even the people developing tools using AI may not fully understand all that it is doing, why it is doing things a certain way, and the tradeoffs. Business leaders are in a tough position of trying to decide what role AI should play in their business and how to balance the risk with the reward. Here are three best practices that can help. 1. Be careful what data you expose to an AI-enabled tool. Uploading your quarterly financial spreadsheet and asking questions to do some analysis might sound innocuous. But think about the implications if that information were to get into the wrong hands. Don't give anything to an AI tool that you don't want an unauthorized user accessing. 2. Validate the tool's output. AI hallucinates, meaning it confidently produces inaccurate responses. There have been numerous media reports and academic articles on the subject. I can point to dozens of examples personally as I've experimented with AI tools. When you ask an AI tool a question, it behooves you to have a notion of what the answer should be. If it's not at all what you expected, ask the question another way and, as an extra precaution, go to another source for validation. 3. Be mindful of which systems your AI-enable tool can hook up to. The opposite side of the first point is that if you have AI-enabled tools operating within your environment you need to be aware of what other systems you're hooking those tools up to, and, in turn, of what those systems have access to. Since AI is a black box, you may not know what is going on behind the scenes, including what the tool is connecting to as it performs its functions. There's a lot of optimism and excitement about the potential upside for enterprises that embrace AI. Fortunately, the past has shown that security is integral to reaping the positive impact of new technologies and processes that are brought into the enterprise. In the rush to capitalize on AI, get ahead of the security risks by committing yourself to understanding the tradeoffs and making informed decisions. EXPERT OPINION BY MARTIN ROESCH, CEO, NETOGRAPHY @MROESCH

A recent AI pivot at an accounting giant has some people thinking that AI's invasion of the job market is well underway.

Fears that AI will be stealing jobs were given fresh life on Wednesday, when accounting giant Intuit announced it would lay off 1,800 employees as part of an AI-centered reorganization. The cuts will affect 10 percent of workers at the company, which owns accounting software TurboTax and QuickBooks. In a memo sent to staff, Intuit CEO Sasan Goodarzi noted that aligning the business with AI will make it competitive as technological change sweeps the economy. "Companies that aren't prepared to take advantage of this AI revolution will fall behind and, over time, will no longer exist," he wrote. The layoffs, which will be completed in September, are not a result of economic hardship, according to Goodarzi, who maintained that Intuit is "in a position of strength" financially. (Laid off employees will receive at least 16 weeks' severance and a minimum of six months' health insurance coverage.) Rather, Goodarzi cited poor performance as the motivating factor in laying off 1,050 of the company's 1,800 employees. "We've significantly raised the bar on our expectations," he wrote. Goodarzi added that the company would replace departing staff at a rate of 1:1 by creating new roles bolstered by generative AI tools. "We will hire approximately 1,800 new people primarily in engineering, product, and customer-facing roles such as sales, customer success, and marketing," the CEO said in the memo. In an email to Inc., a company spokesperson said that the layoffs are "about increasing investment in key growth areas: Gen AI, money movement, mid-market expansion, and international growth." Intuit's shift to AI-oriented labor is happening amid fears that the technology could displace droves of workers. According to a June survey conducted by Duke University and the Federal Reserve Banks of Richmond and Atlanta, two-thirds of the American CFOs who responded said their companies are looking to replace human workers with some kind of automation. Over the last year, 60 percent of the 450 companies surveyed said they have "implemented software, equipment, or technology to automate tasks previously completed by employees." AI software, which is often used to produce text, audio, and images on demand, is increasingly viewed by company leaders as essential to competitiveness. Last August, Erik Brynjolfsson, a professor at the Stanford Institute for Human-Centered AI, spoke to the New York Times about a shift in thinking regarding AI capabilities. "To be brutally honest, we had a hierarchy of things that technology could do, and we felt comfortable saying things like creative work, professional work, emotional intelligence would be hard for machines to ever do," he said. "Now that's all been upended." Additional data shared with Inc. indicates that startups are turning to OpenAI's text generation tool, ChatGPT, in lieu of hiring freelancers on gig work sites, such as Fiverr, Upwork, and Toptal. According to an unpublished survey by the accounting firm Kruze Consulting, the number of startups paying for enterprise versions of ChatGPT has exploded since 2023. Nearly two-thirds of the companies on Kruze's client list of 550 VC-backed startups are paying for the service, Kruze reports, whereas, the average spend on freelance copywriters has plunged by 83 percent since November 2022. "Basically, startups aren't spending money on outsourced marketing -- mainly writing -- now that they can use AI," Healy Jones, VP of financial strategy at Kruze, said in an email to Inc. While recent news and figures make for grim reading for freelance copywriters, data indicate that AI's incursion into other roles has been less dramatic, at least for now. A recent survey by researchers at the Massachusetts Institute of Technology found that companies could only replace 23 percent of wages paid to human workers with AI tools performing the same jobs. Researchers determined this by assessing the current cost of using AI models to perform certain tasks, and then comparing that cost with compensation for human workers. "This is not something where all of the jobs are replaced immediately," Neil Thompson, director of MIT's FutureTech research project, said in a press release last month. Nonetheless, layoffs at high-profile companies like Intuit will signal that a technological tipping point is near -- and along with it, more fallout for workers.

We All Know AI Can't Code, Right?

If anyone is telling you that AI can code what you need coded and build what you need built, they are lying to you. This is not speculation. This is not bombast. This is not a threat. We know enough now about how AI works, and especially GenAI, to be able to say this with confidence. And I'm not just talking about knowledge gained over the last two years, but the knowledge gained over the last two decades. I was there at the beginning. I know. For a lot of you, I'm telling you something you already know as well. But your work here is far from over. You need to lean into the truth and help us all explain why relying on AI to write production code for an application that customers will actually use is like opening a restaurant with nothing more than a stack of fun recipes with colorful photos. They look great on paper, but paper doesn't taste very good. The Boring Structural Work Matters To put this into a perspective that everyone can understand, let me ask you a question: Q: How would you know if this article was written by AI? A: Because it would suck. Yeah, maybe the bots could imitate my vibe, adopt my writing tics, and lean into the rule of threes as I often do, but even then, the jury is still out on how closely it can replicate my style beyond a sentence or two. Banana. Screw you, AI. The thing I'm 100 percent sure AI can't do is take my decades of experience in the topics I choose -- topics that need to be timely across an ever-changing technical and entrepreneurial landscape -- and use my snarky words and questionable turns of phrase to put insightful, actionable thoughts into the heads of the maximum amount of people who would appreciate those thoughts. That's structure. It's foundational. It's boring. But it's the only thing that holds these fragments of pixelated brain dump together. Look, if you want to write about a technical or entrepreneurial topic, you either need to a) spend a lifetime doggedly nerding down those paths with real-world, real-life stakes and consequences, or b) read a bunch of articles written by people who have done just that and then summarize those articles as best you can without understanding half of what those people are actually talking about. Which one sounds more like AI, a) or b)? Now let's talk about how that relates to code, because hopefully you can already see the connection. AI Is Not an Existential Threat Real coders know. The threat that AI presents to your average software developer is not new. Raise your hand if you've ever used GitHub or Stack Overflow or any other kind of example code or library or whatever to help you get started on the foundational solution to the business problem that your code needs to solve. Now, put your hand down if you've never once had to spend hours, sometimes days, tweaking and modifying that sample code a million times over to make it work like you need it to work to solve your unique problem. OK. All of you who put your hands down. Get out of the room. Seriously. Go. We can't have a serious discussion about this. Cheap, flawed, technical-debt-inducing, easily breakable code has been a threat to software developers since they first started letting us kids bang on Basic -- let alone the threat of any technology solution that ends with the word "-shoring". The AI threat just seems existential because of the constant repetition of a few exaggerated truths. That it's "free," that it's "original," and that it "works." Here's why that's going to be a race to failure. Position yourself. "AI" "Can" "Code" That's the most judgy, snarky, douchey section header I've ever written. But in my defense, there's a reason why every word is in quotes. Because this is how the lie propagates. Yes, what we're calling AI today makes an admirable attempt at slapping syntax together in a way that compiles and runs. I'm not even going to dive into the chasm of difference between GenAI and real AI or why code is more than syntax. But I will point to the fact that -- even beyond those quibbles -- we're not at anything I'd call viable yet. Damning words from an IEEE study follow: [ChatGPT has] a success rate ranging from anywhere as poor as 0.66 percent and as good as 89 percent -- depending on the difficulty of the task, the programming language, and a number of other factors. I'll let you determine how "difficulty," "programming language," and "other factors" impacts the success rate. Quotes again. Sorry. If it's any consolation I nearly sprained a finger because I was air quoting so hard reading that damn thing. A conclusion of the study (italics are mine): "ChatGPT has not been exposed yet to new problems and solutions. It lacks the critical thinking skills of a human and can only address problems it has previously encountered." So much like my example of why AI-generated articles suck, if you're trying to solve new problems by inventing new solutions, AI has zero experience with this. OK, all you "ChatGPT-4o-is-Neo" bros can come at me now. But it isn't just the syntax where AI has problems. Aw, AI Came Up With This All by Itself Code in a vacuum is worthless. Every software developer reading this just went, "Yup." Beyond all the limitations that AI exposes when it creates syntax out of "thin air" (or to use the technical term, "other people's code"), deeper problems start to expose themselves when we try to get the results of that code into a customer's hands. Code without design, UI, UX, functional requirements, and business requirements is a classroom exercise in futility. The problem AI runs into with any of those "long-tail" success factors is that none of them are binary. Zero. So, for example, Figma had to temporarily pull back on its AI design feature when it was alleged that its AI is just copying someone else's design. "Just describe what you need, and the feature will provide you with a first draft," is how the company explained it when the feature launched. I can do that without AI. I can do that with cut and paste. Figma blamed poor QA. Which one sounds more true? AI Is Great at a Lot of Things But not elegance. If your code is not infused with a chain of elegance that connects the boring structural-solution work to the customer-facing design and UX, you can still call it "code" if you want to, but it will have all the value of an AI-generated avatar reading aloud AI-generated content over AI-generated images. Have you ever seen that? It'll stab you in the soul. There's a right way to do things and there's a way to do things well, and I'm not naive enough to rail against the notion that sometimes you just can't do both. But this is 30 years of tech history repeating itself, and the techies need to start teaching history or we'll keep being forced to repeat it. So I'd ask my software developer friends to raise your hand if you've ever had to come in and fix someone's poorly structured, often broken, debt-laden, and thoroughly inelegant code. OK. Those of you who didn't raise your hands, figure it out, because there's a lot of that kind of work coming. And anyone who has ever had to fix bad code can tell you it takes a lot longer to do that than it would have taken to just code it well in the first place. EXPERT OPINION BY JOE PROCOPIO, FOUNDER, TEACHINGSTARTUP.COM @JPROCO