Have you seen the news this week about OpenAI and Anthropic calling for a slowdown in AI development? Dario Amodei published an essay warning of “devastating consequences in a matter of months.” Sam Altman also endorsed the position. Researchers at both companies have been posting about extinction risk and the need to pump the brakes. Stock markets responded accordingly, with SoftBank down 10% and chip manufacturers following suit.

My translation: we see this technology not being much more useful than what it is now. AGI is not coming, certainlly not on the timeline we marketed it to you. The returns on continued scaling are troublesome relative to the costs required to train each successive generation. The first company to acknowledge this publicly and slow down the dollar burn rate will incur the wrath of the market.

So, what do you do? You say this big boogeyman technology will end all life on earth and we all need to slow down together; bonus points if you can lobby for this to be included in the national security bucket, because then the US government can bail you out when private investment starts to taper. The same executives who spent years telling us this technology would transform everything are now telling us it’s so dangerous that democratic governments need to step in and help manage it.

There’s the old Turkish proverb: the axe tells the trees that their lives will be easier, and the trees listen because the handle is made of wood. The AI industry has been telling us stories in our own language for years: “Efficiency.” “Transformation.” “Personalization at scale.”

Someone described the industry’s viewpoint toward resource demands this way: the milking shed and the slaughterhouse don’t need to be popular with the cow. First it was priority access to grid capacity and then it was groundwater for cooling. Now we’re looking at coordinated slowdowns that conveniently align with diminishing returns on capital deployment. Whatever serves the companies’ interests gets packaged as tremendous opportunity or existential threat, depending on which version is more useful at the time it bests suits them.

So what’s to do for a fundraising technologist?

This back-and-forth between utopian promise and apocalyptic warning is profit capture in the form (or the guise rather) of a public service, and organizations committed to social good will feel the instinct to shield themselves from it, and rightly so.

I’d like to think my position on this has been consistent: this technology was never going to be the solution that others want [read: need] it to be. The efficiency gains are real in some contexts; competitive pressure exists and adoption isn’t exactly discretionary. But this hype cycle has been driven by people whose interests are served by hype, and the current pivot to doom is driven by those same people whose interests are now served by doom. These folks don’t have your institution, your donors, or your mission in mind. They have themselves, their profits, their reputations, and their investors in mind. I’d call it manipulative altruism if the altruism were even remotely convincing. It’s hilarious and it’s insidious, and it works. We’ve spent years being told we needed to adopt these tools or fall behind, and we’re now being told the tools are potentially civilization ending threats requiring government intervention.

Considerations for Fundraising Organizations

Budget for tools based on demonstrated value, not promised transformation. Those of us who remember Apple’s WWDC announcements about the “new Siri” coming in late 2024 are still waiting. The promised capabilities kept slipping: first to early 2025, then to iOS 18.4, now to this generation of iPhone. The returns available today are the returns you should budget for. Speculative capability improvements are not a sound basis for procurement decisions. When your vendor tells you that their new AI-powered donor flux capacitor will be transformative at some indeterminate quarter on a roadmap remember that Apple couldn’t deliver a functioning personal assistant with virtually unlimited resources and a decade of head start. Your vendor is not going to outpace Apple.

Stay strong in your discipline. The advancement profession has decades of evidence and knowledge about effective fundraising practice, developed through trial and error by people who did the work, read with their eyes, and thought with their brains. AI can consume all the published best practices and training data available, but fundraising has a way of presenting situations that don’t fit neatly into patterns. The donor whose giving history suggests one thing but whose relationship with your institution suggests another or thhe campaign timing decision that requires weighing institutional politics against market conditions. Your years in advancement are what will get you through these times and those situations, not an LLM trained on aggregated data producing responses calibrated to the average case rather than the one in front of you. Fundraising is at its most interesting and exciting in the exceptions, the edge cases. Rather than cultivating the next AI know-it-all, what if we renewed our interest in cultivating experts in their domain? You only get senior software engineers from junior software engineers, and we only get capable fundraising technologists from those entering the profession in this cacophonous technosphere.

Serenity Now

Our industry’s technology budget is limited and even as push budgets iin larger shops push into the hundreds of thousands and more (which sounds like a lot until you realize how little it buys relative to consumer tech expectations) we can’t afford to chase hype cycles driven by companies whose interests diverge from ours. Proven tools, experienced staff, infrastructure that serves our missions rather than vendor valuations are where we will always feel safest, so serenity now–insanity later?