Blake Oliver has a confession that might shock other tax professionals: his CPA license is suspended. Not for anything scandalous—he simply forgot to complete his CPE requirements. The irony isn’t lost on him. He founded Earmark, a platform that allows CPAs to earn continuing education credits by listening to podcasts, and he hosts The Accounting Podcast, which provides 150 hours of instructor-led credit annually.
“I forgot about Arizona’s requirement for 16 hours of in-person CPE,” Oliver admits. “I lost track of it just like everybody does.”
This honest moment sets the tone for a practical conversation about artificial intelligence in the tax profession. In this episode of Federal Tax Updates, hosts Roger Harris, EA, and Annie Schwab, CPA, sit down with Oliver to cut through the AI hype and explore what this technology really means for practitioners today.
The Reality Check Tax Professionals Need
“AI can be overwhelming and slightly intimidating,” Schwab admits, voicing what many professionals feel as they navigate the flood of AI tools and promises. And Oliver agrees.
“It’s kind of janky right now, to be honest,” he says. “It works sometimes, but it doesn’t work a lot of the time.”
He compares the current state of AI to self-driving cars in his hometown of Scottsdale, Arizona. The Waymo vehicles work well enough to get you to the airport, but they take longer routes and can’t go everywhere yet. “Every month, every year, they get a little bit better,” he explains. “It’s always slower than anyone in tech wants to estimate, but we will get there.”
This gradual evolution mirrors what happened with online shopping and cloud computing. Oliver reminds listeners how Amazon started by only selling books, with week-long delivery times. “It happened fast, but also so slowly that you didn’t even notice it happening,” he reflects.
Similar to the early days of the web, AI technology works, but it requires patience and experimentation to unlock its potential. And despite the marketing hype, AI won’t replace tax professionals anytime soon—especially not for the complex, judgment-based work that defines the profession.
Why AI Struggles With Tax Work
Many assume AI will revolutionize tax preparation, but Oliver explains why tax work is actually one of the most challenging applications for current AI technology. And that’s good news for practitioners worried about job security.
“When people think of tax, they think of the administrative stuff.” Oliver observes. “They just think of putting numbers in a form and letting it calculate the end result.” But experienced practitioners know the reality is far more complex.
The tax code is a complex ecosystem of statutes, regulations, court cases, and informal IRS guidance. These layers often conflict, requiring judgment that current AI simply cannot replicate. Harris summarizes the problem perfectly, “AI is great in a world of black and white. But most of what we have to do in our brain in taxes is gray.”
Why AI falls short:
- They pull the first answer they find, not the most accurate.
- They lack real-world experience.
- They cannot apply nuance.
- They forget context as soon as the prompt ends.
Oliver compares today’s AI models to “a very, very smart intern with all the knowledge. They’ve read all the books, but forget everything you say.” This brilliant intern has no real-world experience and needs extremely detailed instructions for everything. Unlike humans who update their understanding through experience, AI models are essentially frozen once trained.
The economic reality is “to actually have an AI that can do tax analysis like you do, Annie, it might cost more than you. Seriously,” Oliver tells Schwab. “That’s why Microsoft wants to buy a nuclear power plant.” The computational power required to handle the multi-layered reasoning and judgment calls that tax professionals make every day would be astronomically expensive.
Practical AI Applications That Actually Work
While AI may not be ready to handle complex tax analysis, it’s already proving valuable in some specific areas. The key is choosing the right tasks.
The sweet spot for current AI technology is “all the administrative minutia that gets in the way of actually getting the numbers into the form,” Oliver suggests.
- Document organization: Document organization is one promising application. “There are tools that will just go through your document storage and look at a folder of client documents they’ve dumped in and just organize them, rename them and put them into your tax software,” he explains. AI can even identify what’s missing and automatically request it from clients.
- Quality control: is another area where AI shines. If you’re comfortable uploading a draft return with source documents, AI can follow your firm’s review checklist and identify issues. “That reviewer is very expensive,” Oliver points out. “And you’re probably reviewing it as the partner in the end anyway.”
- Second opinion: The medical field is another compelling exampleuse case. Oliver describes how British radiologists use AI trained on thousands of cancer scans as an instant second opinion. “The model is actually as good or better than the human,” he notes. The result? Fewer false positives and negatives, and lives saved.
- Challenging logic: Perhaps most intriguingly, Oliver advocates using AI to challenge your thinking. “Give it your analysis and say, ‘Here’s the situation and here’s what I believe. Poke holes in my argument. Be the tax court and challenge me. What am I not thinking of?'”
- Responding to IRS notices: During his presentation at Padgett this summer, Oliver demonstrated AI another practical use: responding to IRS notices. He used AI to draft the response, then had it review his work to ensure nothing would trigger a rejection.
- Research: Schwab shares her experience researching SECURE Act 2.0 legislation. “I feel like it’s learning what’s in it,” she observes. “My results are getting better as more time goes by.”
As Harris notes, if you would hire an intern to do it, AI might handle it better. The goal isn’t to eliminate humans but to free them for work requiring professional judgment.
Getting Started Without Breaking the Bank
With countless AI products flooding the market, Oliver has tips to start experimenting without overspending or compromising security. “Never trust the marketing message or the salespeople all on their own. Trust but verify,” Oliver advises. He recommends starting simply and affordably. Plans start around $30 a month.
Recommended platforms
- ChatGPT Plus
- Claude Pro
- Perplexity Pro
Oliver stresses: Never use free versions with client data. “If you’re using the free version, they can take that data and use it to train the model.”
The real challenge is learning to use the tools effectively. Success requires developing prompt engineering skills—essentially learning to delegate to this brilliant but context-lacking assistant.
“You have to think, if someone was asking me this question, what context would I need to answer it?” Oliver explains. This parallels a skill many tax professionals struggle with: delegation and management.
Tips for effective adoption
- Dedicate 10–20 minutes daily to experimentation.
- Use voice mode rather than typing. This helps overcome years of keyword-based Google search habits, and allows you to retrain your brain to speak to the AI like a real person.
- Compare outputs across platforms before buying niche tools.
- Always keep a human in the loop. Don’t damage your own brand by letting AI speak for you.
The Path Forward for Tax Professionals
“We’ve been through this before,” Harris observes, recalling fears about cloud computing and other trends. Oliver agrees, noting how his podcast was originally called “The Cloud Accounting Podcast” but had to drop “Cloud” because it became the default within five years.
The good news for those feeling behind? “We are just tipping past that point of the vanguard,” Oliver explains. “Now, we’re starting to get to the maturity point where the early majority is going to start adopting it, and that’s going to be like a ten-year process.”
“If you haven’t started yet, you’re not too late. You are still in the first year, you’re in the 90%,” Oliver reassures. “But get moving on it.”
The key to success lies in curiosity and a willingness to experiment. “If you have a curious person in your firm and you give them this tool, they can learn anything so fast,” he observes.
The firms that thrive won’t master every AI tool or automate every process. Instead, they’ll thoughtfully integrate AI to enhance human expertise and free themselves for high-value advisory work.
“Without risk, there is no reward,” Oliver reminds listeners. “You have to be willing to take a little bit of risk to benefit from this new tech.”
The transformation will be gradual enough to accommodate those willing to evolve. As Harris predicts, “A few years from now, we’re going to look back and wonder what the big deal was.”
Start with a $30 monthly subscription. Test it on routine tasks. Use it to organize documents or draft IRS correspondence. Keep the human in the loop. Your professional judgment and ability to navigate tax law’s gray areas remain irreplaceable.
As Oliver’s suspended CPA license reminds us, we’re all human. We forget things, make mistakes, and sometimes struggle with new technology. But that’s precisely why AI can be such a valuable partner—not as a replacement for our expertise, but as a tool that helps us be more effective at what we do best: solving complex problems and serving clients with wisdom that no machine can replicate.
Listen to the full episode above to hear all of Oliver’s insights on navigating AI’s promises and pitfalls in the tax profession.
And, for help with technology, practice management, tax guidance, and more, learn more about having your firm join the Padgett network.