AI, Where to next? Four ideas for institutions, not just Enterprises
Hyperbole and fearmongering aside, AI’s reach into every corner of business and public life will keep accelerating. We’re fast approaching the phase where putting ‘AI’ in bold on a slide deck is enough to pique investor interest, regardless of how rudimentary or absent the actual use of it is.
I want to widen the lens beyond the enterprise to where AI/ML could genuinely help institutions and public life, with the same caveat that always applies. The hard part is assembling a large and diverse enough dataset to train and validate any of it responsibly.
Transparency AI: Imagine a model that traces every major political or corporate decision back to its likely beneficiaries, made visible for public review. For the cautious, that risks analysis-paralysis; for institutions genuinely committed to clean governance, it forces a level of transparency that’s hard to manufacture any other way.
Fact-Checker AI: A trusted, bias-audited source of truth would meaningfully reduce the speed and reach of misinformation and disinformation across social and digital platforms. A service that media outlets (CNN, NYT) and platforms (FB, Twitter) alike would plausibly pay to subscribe to, provided the model’s own training and incentives are open to scrutiny. The same approach extends to past news, into disclosing who funded a given study before its conclusion shapes public behavior.
Influence-Mapping AI: Regardless of where you sit politically, the pattern of retired officials moving into advisory roles that bridge industry and policymaking is a real and traceable phenomenon. A model that surfaces the linkages and funding trails behind policy influence, transparently and without picking a side, would be a genuine public good. The value is in the visibility, not in the verdict.
Equity-Gap AI: A model that simply surfaces where policy treats a small business differently than a large one for functionally the same reason, a handful of jobs promised versus thousands, without editorializing on which treatment is correct, would at least make the tradeoff visible enough for voters and leaders to debate it on the merits. The same can be extended to taxation policies and other policy decisions.
As I said in the first post of this series, AI has excited me the way the early internet did in 1995. I lean optimist, and I expect the constructive uses to outweigh the destructive ones over time. Design thinking applies to every one of these ideas. If anyone wants to take one on, I’d like to be part of that effort and I’d like to see it built firsthand.