Government communicators get ready for another boot off the bus
UNSPLASH PHOTO
It is a saying among professional communicators, whether in the public or private sector: they are the last on the bus and the first to be thrown under it.
Communicators are usually brought in at a late stage of an initiative that has been months in development, and if something goes wrong, they are the first in line for blame for allegedly not communicating the initiative properly.
When it comes to government spending cuts, like those about which there have been growing rumblings around Ottawa this summer, communicators—those who draft press releases, speeches, communication plans, manage our media relations, and run government social media accounts—are among the first to get thrown off the bus.
This is because while they are essential to a government's public image, communicators do not deliver public services. During the last serious attempt to rein in federal spending — the 2012 Deficit Reduction Action Plan (DRAP) under Prime Minister Stephen Harper — communicators were lumped into a category of employment referred to as “back office,” a term that did not radiate importance at the time.
While federal communications staff are not classified as a distinct employment category in the federal government, the category they are captured under, Internal Services, ended up accounting for upwards of 40 per cent in staff cuts, depending on how you measure it.
Prime Minister Carney’s expenditure review exercise continues to gain momentum, and communicators are bracing for another hit. But this time around, they have a new reason for worry: the apparent writing skill of the artificial intelligence large language models that are now in wide public use.
Anyone who has ever used Chat GPT, for example, can tell you that it can write clearly, fluently, and succinctly. On the other hand, anyone who has ever worked in government communications, as I did for many years, can confirm that there is a lot of bad writing going on at the communications working level. As a communications executive, I routinely had to spend hours massively revising or rewriting products completely. I once received a communications plan that had 27 key messages, whereas our standard practices required that they be limited to four.
The reasons for this are many and varied. Lawyers and policy leaders usually have the final word on a text, which can make it ponderous and hard to understand. But whatever the cause, the writing that crossed my desk was frequently awful. My working life would have been made immeasurably easier had I received drafts that only required tweaks for me to be satisfied and approve.
When you consider that accelerating the adoption of AI to increase government efficiency is a priority for the Carney government, harnessing a secure version of something like ChatGPT to improve the quality of government communications writing would seem like low-hanging AI fruit. It might be sorely tempting at the senior levels of communications branches to champion the use of secure AI as a low-cost alternative to whole categories of junior and middle-level writing positions that, in practice, add little value to finished products. And the savings could be permanent because the re-hires that often occurred in communication branches after DRAP, for example, would be unnecessary.
Communications branches might be the obvious choice for early adoption of AI for writing purposes; however, it could be applied throughout the federal government to thin out junior and mid-level categories with significant writing responsibilities, of which there are many thousands. My experience is that the quality of writing of many of my colleagues was poor and that it tended not to improve over time. But the evidence is that AI models actually do improve as they learn more. That can't help but be a temptation to cost-cutters.
Canadian understanding of and trust in AI is among the lowest in the world. A recent KPMG study ranked Canada 44th out of 47 countries in AI literacy and 42nd out of 47 in terms of trust. This, and the searing memory within the civil service about the Phoenix pay debacle, will make a rapid embrace of AI by the federal government a scary, uncertain proposition.
Artificial Intelligence and Digital Innovation Minister Evan Solomon has formulated an AI adoption strategy that emphasizes, among other objectives, scaling domestic champions and building trust. He will want to avoid another big federal IT swing and a miss. He could help achieve either of these objectives by pushing for AI adoption to improve federal government writing, in communication, and beyond.
So, if you are a federal communications writer, this is going to be a tough summer. You know you will likely be on the front lines of those who get bounced from the bus because you always have been. But now you may also have to worry about being replaced by a large language model.
That would be hard to take. But at least the writing might be better.
