Note: this is a surprise bonus post in a blog series about the relationship between Oracle and AI, specifically in the EPM space.

A Glass Sculpture with a Building in the Background by Allison Saeng from Unsplash

When I launched my 5^th and final blog post in the Oracle + AI series in April of this year, I was pleasantly surprised to see an email within 24 hours from my buddy and former CEO, Edward Roske. Edward and I share a deep fascination with AI.

We chatted over email about some of the findings in my testing with the free version of ChatGPT. Edward sent over some results from the paid version of ChatGPT, which were lightyears better than the output from my free testing. As we delved deeper into the conversation, I realized a bonus blog post was in order. The insights Edward has into GenAI are interesting and thought provoking, which should not surprise anyone.

So here it is, in interview style, a chat with the one and only Edward Roske.

Quick introduction to Edward Roske: Edward founded interRel Consulting over 25 years ago, a firm focused on Oracle EPM and its predecessor, Hyperion. interRel became a formidable boutique firm in the Oracle EPM space. interRel joined the growing ranks of Argano back in 2020.

Edward left Argano in August 2023 to pursue a career in GenAI. Since earning a data science degree 2 years ago, he’s realized how important GenAI has become and how irrelevant his degree has become in such a short time. Edward believes we will see more change in the next 3 years with AI than we have with any other software technology in the last decade. Edward currently serves on the ODTUG Board.

What have you been up to these days?

Since leaving Argano, Edward has focused entirely on AI. He spent the first 3 months devoting himself to learning it. He’s changed the trajectory of his career. 

First, Edward and a friend of his from Argentina cofounded a company that provides AI-based therapy. Called GetsMe.ai, this is an AI therapist and has a free starter plan. This business takes up a large portion of his time.

AI therapy? Sounds intriguing. Why?

Edward became concerned about his kids, the current mental health crisis, and the rising levels of anxiety and suicide in GenZ (ages 15-29). Edward decided he wanted to give something back.

He also became an investor and co-owner in an AI services company that helps AI improve, automate, and optimize back-office functions.

Finally, Edward is teaching and guest lecturing at a few universities on the topic of AI and the office of the CFO.

You and I both have been obsessed with GenAI for years now. Talk to me more about your journey with GenAI—why were you attracted to this technology?

“It’s a transformative technology. AI has been making exponential baby steps and accelerating at a super-fast rate.” He says it’s not linear but exponential growth, which is faster than any other technology he’s seen.

“We are in a world now where if you can conceive of the question, you can get the answer. If you were to ever ask yourself the question ‘can AI do it?’ the answer is yes. People are thinking entirely too small with it.”

Edward believes there are 3 stages for any new technology:

• Stage 1: can it replace some of the costs of what we’re doing today?
• Stage 2: can it replace entire functions because the technology can do them as well or better than a human?
• Stage 3: what problems can it solve that humans never could?

He sees the potential of the 3rd stage of GenAI. For instance, can AI do things that an EPM consultant could never do? Sure, it can create a calc script and with only a little teaching. However, “eventually, it’s going to be able to do things with planning and forecasting that we humans can’t even conceive of.”

Edward believes he’s gotten good at filtering out what’s hype vs. what’s real. “GenAI is not a hype. We’re all underselling it. It has the potential to affect every aspect of our lives.”

How do you use GenAI in your everyday work life? Is there any EPM overlap in your new career focus?

The #1 way he uses AI in his life is as an assistant. He uses it for copyediting, etc.

However, at the beginning of his AI journey, he first researched how GenAI could help the EPM world. EPM is now a transitional part of his life, yet it’s where his real-world experience is. Although he stays current on EPM, he’s curious how GenAI will disrupt the EPM world.

Always a disruptor himself, Edward dares to challenge the status quo. “Do we still need EPM consultants? Do we still need EPM contractors? There’s a disruptive component to GenAI. All EPM vendors try to mention the word ‘AI’ as much as possible to look relevant. No one has said if we built EPM in 2024 what would it look like? Is a grid the right way of doing it? Should we be planning the way we’re planning today?”

You introduced me to the GenAI tool Claude, and it seems to have a leg up on ChatGPT with its real-time abilities. That begs the question in a world where everyone knows ChatGPT…what GenAI tools shouldfolks be evaluating?

One important note from Edward: “[most of the] current GenAI tools out there cost ~$20/month.” Edward fully believes everyone needs to be using a version of AI. “Someone using AI will take your job if you’re not using it.”

First, he mentions that some GenAI tools are more corporate-minded, and some are more personal-minded.

On the personal-minded side, he believes there are 4 big names:

1. ChatGPT: with 100+ million active users per week, most people are using the free version. The free version is currently based on GPT 3.5. The previous paid version is currently based on GPT 4. There is a huge difference between the two. As Edward mentioned, “GPT 3.5 passed the bar in the bottom 10% and GPT 4 passed the bar in the top 10%. The paid version, ChatGPT+ is just so much better.”

Real-time news release from Edward: as of May 13 (this week!) “everyone now has access to the best possible version of ChatGPT. The model is called GPT-4o (‘o’ for ‘omega’) and it’s real-time and multi-modal.” He says to think of this version as the movie Her except everyone gets access. You no longer have to pay $20/month unless you have higher than normal levels of usage. Exciting stuff!

• Microsoft Bing has a Creative Mode and, under the covers, it’s the free version of ChatGPT 4. It sells under the name Copilot. 
• If you want a big context window, use Google Deepmind’s Gemini Pro. A context window allows you to feed the GenAI tool reference material, giving it context and direction on the response you’re looking for.
• Anthropic Claude Opus: “it’s better than GPT 4 because it’s better at esoteric coding. It’s also better at being human. It has flashes of a “ghost in the machine’. Use the pro version.”

On the corporate-minded side, “Microsoft is really leading the way with Copilot. It’s in Windows 11. The #1 use case for it now is with coders.” Remember that GitHub is owned by Microsoft. Microsoft also owns half of Open AI. 

What do you recommend in the GenAI space as it relates to EPM?

As Edward observed, “most companies getting into the AI scene are applying AI on top of their offerings. Or they’re creating a model like what Amazon does with AWS or Microsoft is doing with Azure.” Microsoft is weaving it into their core fabric. “Are there any real EPM options out there? Other vendors are trying to figure out how to just put it on top of what they already have.”

Edward believes Microsoft’s Copilot is a serious competitor to any EPM tool out there today. Microsoft offers two flavors of Copilot. First, there’s Copilot for Office. It costs $30/user/month. With Copilot, Edward can open a blank PPT and create a starter PPT based on a prompt. “You can also use it to write an email to someone. Or go into Word and have it brainstorm an outline for an essay.” It’s his new favorite friend. As he joked “I don’t have any real friends, just artificial friends.” Suuuuuure, Edward. 😊

There’s also Microsoft Copilot for Finance. “It mostly lives in Excel and ties into your ERP.” Example questions you can ask it:

• Can you analyze our financials and tell me what’s missing budget by a lot?
• Can you tell me if there are any missing transactions?

What concerns do you have about AI in general? Do you think the singularity is nigh?

We both chuckled here. As with any new technology disruptor, AI comes with risk. Edward has 3 main concerns about AI:

1. Job displacement. “AI could be the reason why we all need universal basic income. There’s a realistic possibility—job loss, job displacement. However, every time a new technology comes in, people are freed up to do more impactful things and we end up needing more people instead. Temporary displacement and long-term job creation occurs. Like the movie Hidden Figures, human calculators were replaced by computers, but in the long-term so many jobs were created by computers. It just increases access to more people.”
2. Bad actors. “Anthropic Claude is constitutional AI. They give their AI a set of values and beliefs. No one else creates their AI this way. Anthropic will think about what it’s been asked. On the other hand, GPT 4 knows how to build a nuclear weapon—it’s just been told that it’s not allowed to do it. Bad actors could use AI to make people more evil.”
3. Safety. Is AI going to kill us all? He’s not that concerned about this one. He has faith in the people who develop AI. “There isn’t just one Skynet but millions. Their threats are themselves.” He said jokingly that he believes there’s a 0.2% chance that it’s going to kill us all. “Humans will probably kill themselves first.”

What are you most excited about regarding the future of the AI space?

“I’m really excited about how much time this is going to free up for humans to go after all the things we never get to. To answer all those questions we never get to. This will get people out of the boring grunt work and free us up to actually use the [school] degrees we got.”

“This is going to solve problems that we can’t even imagine right now.” He believes AI can already answer questions that we can’t conceive of. For instance, Edward asked GPT 4 and Claude what the patent looks like for time travel. He estimates what they designed is 80% of the way there. “The models we’re using today will be the worst version. The doubling capacity for AI is every 5-14 months, which is faster than Moore’s law (which doubles every 2 years).”

Give us a preview of your AI sessions at ODTUG Kscope24.

Edward has two sessions at Kscope regarding Oracle EPM and AI: “Advancing Oracle EPM and Analytics with AI: A New Frontier in Data Analysis“ and “Empowering Oracle EPM with ChatGPT: Mastering Groovy Scripting”. In his Groovy session, he’ll cover how to use ChatGPT to write Groovy scripts. “No one should be learning coding right now. Use ChatGPT to replace the coding stuff that we don’t need to know how to do.”

He’s also moderating an Oracle-AI panel: “The AI and Oracle Roundtable: Insights and Interactions.”

Finally, Edward is leading Kscope’s new AI day which is on Monday.

A big thank you to Edward Roske for continuing our conversation about all things GenAI and volunteering your valuable time. We’ll both see everyone at ODTUG Kscope24!

The Full Oracle+AI Blog Series

Did you miss a post? Here are all of the posts in this Oracle + AI blog series:

• Oracle + AI Series: My Superhero Avatar
• Oracle + AI Series: Continuous Innovation with EPM Apps
• Oracle + AI Series: Analytics + AI = A Data-Driven Marriage
• Oracle + AI Series: Next Gen Oracle EPM Reporting
• Oracle + AI Series: Revitalizing Your Oracle EPM Career with Personal GenAI
• Surprise bonus post! Oracle + AI Series: GenAI in the Wild (this one)