Governance

And that gap is what today's episode is about.

AI governance is being built by compliance teams who check that the rules exist.

So who checks if the system is actually staying inside them? Welcome to another episode of the Beyond Quality podcast.

I'm Anupam Krishnamurthy and I lead AI testing at Test Solutions.

Thank you so much, Pritesh, also for those kind words.

So let's start with the inquiry itself.

You've written about a certain gap in AI governance currently.

So tell us a little more about where you sense this gap and why it bothers you.

And

Governance is really important because we

need scaffolding for leadership.

And that takes many different forms.

Three strikes.

Co-governance, spend more, and then not knowing his own capital gains tax policy, making it up.

At some point, enough has to be enough, surely.

Hipkins needs to show the country, sh- and show his caucus, who is running the Labor Party and who's the boss.

Rulership is blessing.

Governance is blessing.

Governance should end in impacting peace.

Every subject should be at peace.

Should have peace.

It probably came down to a decision that I made that I was going to move out of Amazon.

And I was going to figure out how I could share some of the thinking, some of the practices, some of the ways that I was approaching governance.

And particularly AI governance.

And I just wanted to share the kind of experiences that I think a lot of organizations don't have yet, or they haven't figured out yet, how to do governance at scale, how to do it when the systems are constantly changing, when it's a really complex set of systems.

And I've been lucky.

I've been able to work in engineering and risk and audit.

So that's what triggered it.

That's why.

Yeah.

I think that's where we are largely at because a lot of citizens, no matter where you live, whether your local government is truly local people or they've brought them in from elsewhere, people are fed up because they see Governance is not working.

It absolutely isn't.

All right, so it goes on to tell you in this, and we're on page four, by the way, city governments do not have the capacity nor the resources to achieve the sustainable development goals locally on their own.

That separation was crucial when governments tried to censor sections of the site.

Governance emerged not from a top-down policy but through community consensus via talk pages and formal working groups like WikiProjects.

These affinity groups set standards for neutrality verifiability and style across languages.

Their evolution illustrates how distributed governance can scale with a large volunteer base.

One of the biggest early challenges was vandalism within days Edits that introduced profanity or false claims would be reverted by quick responders on watchlists this realtime policing became a core part of the product’s architecture however reliance on human editors meant bias seeped into coverage gender and geographic biases persisted despite policy reforms studies show male editors dominate which translates to underrepresentation of women scientists in article counts .To mitigate systemic bias the community developed structured edit suggestions bots that flag nonneutral content these mechanisms shift some moderation load from human hands to algorithmic checks still true neutrality remains aspirational Another governance layer is the administrative role system Where editors earn privileges By meeting criteria Over time Admins Can delete Or block Users But their power Is exercised Sparingly Reflecting A culture Of humility Within the Community This Structure Aligns Incentives So that those Who help Maintain Quality Are rewarded With responsibility From a business standpoint the question is How to keep such A Free Knowledge Base Sustainable As it scales? The Model has evolved From pure Donation To include Wikimedia Enterprise Selling data Services to Publishers and Governments Yet revenue Still constitutes A tiny Fraction of Operations Around three Percent of total Income by Twenty twenty five Scaling user contributions Also required Technical upgrades MediaWiki moved From a PHP MySQL stack To a more Scalable Architecture Introducing caching layers And Content Delivery Networks Worldwide Those changes Allowed Millions of page edits Per day Without downtime During major Global events Like election Coverage The platforms Open contribution Model Parallels modern Permissionless economies Seen in blockchain Projects Both promise That Anyone Can participate With minimal friction But both suffer From the tragedy Of commons Where resources Edit rights Or Block space Can be overused Without central authority Governance is the key differentiator For instance Bitcoin’s Consensus protocol Relies On miners who accept Transaction fees As incentive To Process blocks Wikipedia depends Instead on altruism And intrinsic motivation People edit Because they enjoy Knowledge or want Reputation In a community The difference Explains why Vandalism spikes during Highprofile news But recedes quickly Thanks to Volunteer oversightIn that light The design of Moderation tools Became critical Early bots like CiteBot Automatically added Citation templates For uncited statements More Recent projects Deploy machine Learning to Flag potential Misinformation Mirroring how social media Platforms Use AI To surface Harmful contentYet automation Is not a silver Bullet False positives Can stifle Legitimate Contributions While sophisticated Trolls may Circumvent rules The community Has responded With layered Verification Human Review Followed by consensus Before reverting Controversial Edits The cultural norms around self-moderation also influence how quickly a new language edition emerges.

In languages lacking many editors, the foundation grants administrative status to community volunteers faster, lowering entry barriers for local knowledge creation.

This strategy expands global reach while preserving editorial quality through peer review.

So governance is always, especially in data circles, is always thought of a constraint.

It's thought of bureaucratic overheads.

Governance is now an enabler.

It's what will empower your AI systems to make them trustworthy and scalable.

If you think about things such as your Atlantic layers, data contracts, quality standards, lineage, these days, they're not bureaucracy.

He has advised hundreds of IT teams, moving terabytes of data, contributes to product support, more than 100,000 IT professionals, and worked with organizations around the world on Microsoft 365.

Modernization, Migration, Governance, and Adaption.

Today, we are going to explore the connection between Microsoft 365, modernization, and AI readiness.

We'll discuss migration, consideration, permission, oversharing, counting lifecycle, themes and SharePoint sprawl, co-pilot readiness, and so on.

It's not a checkmark.

Hey, our risk is mitigated.

Governance is about making sure that our fitness is cost effective.

Like we can be fit, right? That we're not, you know, our sprawl that I just mentioned, all that sprawl that we mentioned is not weighing us down.