AI chat-fest could up-end community consultation

Nobody likes being ordered around. Government is, at its most fundamental level, unavoidably about just that – laying down the law, ordering people around. Consultation is what governments rely on to take the curse off this. Just about every government decision is preceded by a call for public comment, and every decision is accompanied by a list of the people and organisations who have made submissions. It’s not quite, but nearly, a constitutional convention.

There are problems with this. The people who have monetary interests involved in a decision are much more likely to chip in, which means that rich people are more likely to comment than poor people, underlining power imbalances. People living next door to, say, a road widening are more likely to comment than people who will benefit by a couple of minutes off their commute, distorting cost-benefit analysis. And, of course, the government isn’t obliged to listen to anybody, and indeed the one constant in all of this is that after the consultation process has closed everyone on the losing side will complain that the consultation was only bureaucratic box-ticking and not genuine.

As it happens, both the advantages and the drawbacks of consultation processes have just been explored in detail in a recent Victorian parliamentary report (Legislative Council Environment and Planning Committee, Community Consultation Practices Inquiry, March 2026). This inquiry, just as a data point, received 133 submissions.

As a government inquiry, this report naturally focused on what the government could do better. As it noted (Finding 20), “Poorly coordinated engagement can damage trust in engagement processes, as people can perceive them as being rushed, unclear or unfair.” Its recommendations press for governments to provide just the right amount of necessary information, to reach out to all the people who want to be heard and none of those who don’t, and to have the maximum amount of consultation that won’t overload an overworked and impatient citizenry. All good things, certainly, but with one notable blind spot.

I’m old enough to remember an internet that worked on trust. It’s hard to believe, but back then email didn’t need to have spam filters. You could, for one brief shining moment, contact a roster of the great and famous without impediment. And then came the first Nigerian prince scam letter; and from that point on, the downside of free communication proved to be the ease with which it could be subverted to bad ends.

Today something between 50 per cent and 80 per cent of email is calculated to be spam of one kind or another, and every application has to incorporate a huge range of protections against fraud, exploitation and theft. This sort of works, most of the time; and we may look back on today, too, as a golden age.

Artificial intelligence (AI) is making immense strides, fuelled by unimaginably enormous investments. The jury is still out on whether this will end with robot overlords exterminating us fleshies, but we’re certainly at a point where a large number of things that once required elite bureaucratic and cultural skills can now be handled quite satisfactorily by cheap AI apps. Among those things are making submissions to government. That’s a good thing, at the first level, making it easier for ordinary citizens to talk to bureaucrats in their own language, but anything that makes life easier for me or you makes it ten times easier for people whose actual job it is. Which means (eggs=eggs) that scams, too, will become easier and cheaper.