Marketers who plan for either AI-driven extinction or AI-driven effortless abundance are optimizing for the wrong outcome, argues Greg Jarboe in a new Search Engine Journal Show conversation with Katie Morton. His probability estimate on whether AI will kill us all: "Maybe, but probably not." The real work, he says, is happening in the space between those two headline-friendly extremes — a space that "doesn't get much airtime because it doesn't sell headlines."

The discussion, published this week on Search Engine Journal, spans ten segments covering AI risk, PR strategy behind AI panic, whether AI investment is outpacing usefulness, and how marketers should prepare for the shift. Jarboe's central argument is pragmatic: reality will be less cinematic than both the doom and utopia stories, delivering "real progress next to real problems, some measurable wins mixed in with a fair amount of disappointment."

The Y2K Playbook Still Applies

Jarboe reaches back to Y2K for a working model of how AI warnings should function. Planes stayed in the air and the power grid held on January 1, 2000, but that calm outcome resulted from years of unglamorous fixing — not proof the warning was empty. AI warnings work the same way, he says, when they point at a specific risk and lead to a specific safeguard. They stop working when the message is simply that the sky is falling and nothing can be done.

The opposite pitch — that AI will erase drudgery and hand marketers a golden age — comes from "the same hype machine wearing a different costume," according to Jarboe. Marketers now have to live in that uncomfortable middle instead of picking a side.

Fact-Checking Isn't the Bar Anymore

For SEO practitioners, the more operational point comes next. Most SEOs already know AI hallucinates. Jarboe argues that knowing this is no longer enough, because fact-checking only tells you whether a claim is technically accurate — not whether the claim holds up.

His proposed replacement is ground-truthing. Before shipping anything an AI drafted, practitioners should ask three questions: whether the underlying inference is right, whether the recommendation fits your market and your customer, and what evidence would prove the central assumption wrong. An AI system can produce a clean-looking draft or a fast dataset summary. It cannot tell you whether the direction is correct. "That's still your job," Jarboe says.

Silos Are a Citation Risk Now

The part practitioners underrate, in Jarboe's view, is organizational. AI didn't just change how content gets made — it raised the cost of departments that don't talk to each other. When a press release says one thing, the landing page says another, and social says a third, human readers get confused, and so does every AI system trying to summarize the brand.

"Inconsistency used to be a minor annoyance. Now it's a citation risk," Jarboe says. His prescription: SEO needs to be in the room with PR, and analytics needs to be involved from the start of a project rather than summoned at the end to bless a decision already made.

What to Actually Do This Week

Jarboe puts forward three concrete actions:

• Before publishing anything AI-assisted, run it through the ground-truthing questions rather than a spellcheck-level review. Ask what would have to be true for the conclusion to be wrong.
• Pull one person each from PR, content, and analytics into the next planning conversation for any piece touching brand messaging, even if it feels like overkill. The inconsistencies AI systems pick up on are usually the same inconsistencies departments created independently.
• When PR pitches a story, build the evidence in from the start. A strong hook plus real data and a sample size that survives scrutiny beats a hook alone — and it is "the difference between a piece that gets cited and one that gets ignored by the next model update."

Why the Extremes Dominate

Jarboe attributes the doom-or-utopia framing to structural incentives rather than bad faith. Fewer journalists are covering more AI claims, and the incentives — algorithmic and human — reward the most dramatic version of any story. "Increasingly, your hair has to be on fire to get attention," he says. That pressure strips out the qualifiers, the "yes, but" and the "only if," until readers choose between two extremes that were never the real choice.

His fix is not to abandon the strong hook. It is to back it with a story that can survive an editor's second look.

Search Engine Journal is also hosting a related webinar, "A New Place to Look: Where Your Next AI Citations & Clicks Come From," featuring Lisa Salvatore, Sr. Manager of Integrated Marketing at CallTrackingMetrics, and Brian Barranger, Sr. Account Executive III, covering AEO insights, FAQ content, customer phrasing, and what a qualified conversion sounds like.

For teams tracking how AI systems decide what to cite, the signal to monitor is consistency: whether brand messaging held steady across channels before the next model update either rewards it with a citation or quietly drops it.