The First Woke Wars were 15 years ago. Somehow, we’re still only cancelling the little fish

There’s a term being used to describe the political correctness wars of the 2010s: Woke 1.0. I prefer The First Woke Wars. It’s the period when Black Lives Matter and #MeToo and DEI initiatives all converged to charge valiantly against misogyny, racism, bias, sexual assault, and all the systems that were taking too long to improve. 

There were some wins (Harvey Weinstein, marriage equality), but also a lot of self-inflicted wounds — public cancellations of people that, in retrospect, reflected badly on those doing the cancelling. It’s fine; we all got caught up in it. I’m not going to pretend I didn’t add my voice to baying mobs on Twitter, demanding the cancelling of someone for some perceived slight based on a moral values package we’d all signed up to like a subscription service, and were all too scared to cancel when the content changed. 

Some people of that era now see it for what it became: a wild trench charge with rainbow bayonets affixed, skewering anyone and everyone we could get within stabbing range of. There were righteous battles, but there are also lessons we should learn so that the next wars are fought with more precision and less collateral damage. 

Related Article Block Placeholder Article ID: 1267971

For me, a moment of reconsideration was when they started coming for the comedians. Sarah Silverman losing work because of a poorly thought-out blackface sketch from years before, for example. Or Kevin Hart making daft homophobic jokes on Twitter a decade earlier. It felt like lobbing off low-hanging fruit.

Entertainers don’t have much power or authority (or rather they didn’t back then — the Joe Rogan influence on power hadn’t yet fully become evident). What comedians do is try to find the most surprising/shocking thing to say which will get you to laugh. Getting that wrong is a part of the risk of being a comedian, and finding out the times have changed and your old material has gone from punching up to down is a standard part of the job. I began to feel us stand-up comedians were easy targets to shoot down once people realised those with real power and authority were impervious to the simple weapons we had access to in those early days of Woke. You can’t cancel a country’s government. Corporations don’t buckle under social media pile-ons.  

That’s what I think of when I hear about yet another writer having to defend themselves against claims of AI use: this week’s allegation (and it is becoming a weekly issue) is against Zimbabwean writer Simbarashe Kundizeza, who won a short story competition. The Atlantic says his entry scored 100% on Pangram, that AI-detection tool everyone is putting so much blind faith in that the inevitable fall from grace is more certain than any score it generates. 

Maybe the accusation is accurate. Perhaps yet another non-white writer used AI to generate a few thousand words about grief, then submitted them to a literary prize that only a few people had heard of, and won enough money to cover about a minute’s worth of a data centre’s operational cost. If they did, what a waste of their own time. All the writer had to do was instead use a chatbot to organise the machinery of state power. Then they’d get a contract, a ministerial handshake, and a press release about innovation, not international news coverage dismissing the provenance of their story because an AI said they used AI, resulting in shame and shunning for years to come.

In February, OpenAI announced it was bringing ChatGPT to GenAI.mil, the generative AI platform used by the US military for use by three million civil servants and military personnel. The chatbot’s listed uses for the military include summarising policy documents, drafting and reviewing procurement material, generating internal reports and supporting planning and administrative work. These are writing tasks. For those not paying attention, writing is the thing that writers, who barely make any money, now have to prove they did themselves. All while a company helping the military write its documents gets to establish the norms.

The American Civil Liberties Union has documented the role of Palantir — a data integration, analytics, and artificial intelligence company — in US immigration enforcement, drawing on agency records and other reporting. The union’s account includes the use of tools for identifying and prioritising people for potential deportation, assembling personal information across databases, and using generative AI to summarise, categorise and translate tips sent to ICE. It raises concerns about transparency and the criteria underlying these systems. 

Related Article Block Placeholder Article ID: 1268035

An AI short story wastes an afternoon, yet we flog those caught out for allegedly “writing” one in the town square. Meanwhile, AI inference in an enforcement process can put somebody’s freedom at risk, and governments actively pursue this with glee.

Here in Australia, David Pocock raised concerns that early advice from parliament’s AI committee to Anthony Albanese had been “heavily vetted” and “edited” by an adviser in the Prime Minister’s Office before it was sent to the prime minister. The committee’s Labor chair, Jo Briskey, acknowledged engaging with the office, although the committee leadership denies interference. The advice in question includes reviewing copyright arrangements, which are recommendations about the conditions under which AI companies can use other people’s work.

That deserves at least as much curiosity as the authenticity of a short story. If a writer must produce drafts to explain how their work came into existence, the government can explain its contribution to advice about the industry it is supposed to regulate. Who changed what? Whose concerns made it into the recommendations? Which were left out? And while we’re demanding disclosure, which tools are government agencies themselves using to produce their work? Politicians deciding how much control we retain over our writing should expect to account for their own.

OpenAI and Anthropic have sought changes to Australian arrangements for model training. Creative industry organisations have argued that weakening existing protections could give AI companies access to work without adequate consent and payment. While the government has ruled out a blanket exception to copyright law for AI companies to train on Australian content, it is open to allowing models to train by scraping online content by default, unless creators opt out. It reeks of their approach to gambling advertising. The onus is put on the consumer/creator, not on the corruptor. 

It would be quite an achievement to build a culture in which an individual writer’s suspected use of AI merits a public trial, while the industry’s demand to use writers’ work becomes a technical conversation about investment settings. Authors are now required to demonstrate that the work is theirs. Having done so, they may still have to fight to retain control over what happens to it. 

These, unfortunately, are battles too large to fight. Or so it feels. So, instead, we focus on the allegedly dishonest prize entrant. That’s a much easier story to tell. A writer can be embarrassed, disqualified, publicly shamed. There’s a visible punishment available even if the tools of the accusation are of questionable quality. A multinational technology company embedded in government systems is harder to confront. Its relationships involve contracts, procurement rules, security claims and politicians eager to announce investment. Nobody has to produce a villainous motive for the imbalance to develop. One dispute is much easier to personalise.

Related Article Block Placeholder Article ID: 1267731

The ethics we’re asking writers to adhere to should also apply to governments. If disclosure is essential, tell us where government agencies use AI and what they do with it. Explain whose information and creative work the systems use, and on what authority. Identify who answers when something goes wrong. If we require a writer to explain their process, we can require an institution exercising power over us to explain its own.

A tool that helps with administration or research should be able to demonstrate its benefits and safeguards. The presence of something worth keeping makes it more necessary to establish limits. There’s no shortage of people investigating these issues, including researchers, journalists and civil liberties groups. They’re doing difficult work that needs public attention and political support. The problem is how easily a dispute about a writer comes to stand in for the entire moral challenge of AI, allowing us to feel that an important boundary has been defended once someone loses a prize.

If you want writers to show their drafts, you can at least ask ministers to show their contracts.