I still remember the very first email I sent. My best friend Dave and I each installed the CD-ROM. We set up AOL accounts and created usernames. We called each other in the afternoon via land lines to exchange our “email addresses,” whatever that was. Then I sent Dave an email to meet me that night at our favorite local movie theater at 7 p.m. to catch a film. Then I waited. The suspense was palpable! Would Dave actually be there? Later that night, when Dave walked through the doors of the cinema, we were both stunned. The internet was a digital miracle.
I suspect we’ll look back on Nov. 30, 2022, as a similarly seismic moment when a shift occurred in the relationship between people and artificial intelligence. Work that’s been going on for years behind the scenes to develop ever more powerful AI applications suddenly became public, accessible and practical with the Thanksgiving weekend release of ChatGPT-3.
In the months that have followed, the rate of improvement and innovation feels exponential, with stunning new capabilities announced weekly. Some technology leaders have even called for a six-month “pause” on advanced AI research.
The question isn’t: Is this big? The question is: How big? Is this a disruption equal to the introduction of the iPhone, which made information portable and personalized? Is this big like the release of Google search, which “organized the web” and made a world of online information searchable and accessible? Is this big on the scale of the internet itself, and the massive disruption which moved us from an industrial to a digital economy and society?
I think the answer is yes. This is transformational, not iterative, change. Here are five practices I’m following to lean into this disruption.
1. Avoid either/or extremes
The first reaction I recall seeing from industry peers when ChatGPT was released was, “This is going to put journalists out of a job!” I’d argue a failure to develop sustainable local business models is a bigger immediate threat. In the long run, ChatGPT and its AI peers pose the biggest threat to journalists who don’t explore how to use these tools to enhance their reporting.
Likewise, the explosion of overnight AI-expert social media accounts suggests all the pitfalls of previous “Gold Rush” moments.
So let’s start by resisting the either/or extremes of both the doomists and the magical thinkers. It’s neither the best of times, nor the worst. More likely, it will be a mix of both.
Transformational change will be complicated. Starting with that expectation will help leaders navigate the threats and dangers while finding smart ways to apply AI to problems we have today and to new opportunities in the future.
The old adage applies: Technology is neither good nor bad, but the people who use it are capable of both. So let’s embrace and adopt uses that are beneficial while managing the threats and risks.
2. Start small – but start!
As of this writing, ChatGPT is free, and ChatGPT-Plus is $20/mo. I know people who already use the tool daily. They incorporate it in their workflow as a check, a sounding board, or a second opinion in all kinds of activities from email responses to writing summaries of longer reports to cover letter first drafts.
I wouldn’t stay up until midnight every night going down ChatGPT rabbit holes; but I also wouldn’t “wait for the fuss to blow over” – because it won’t. There are so many simple ways to integrate these new capabilities as everyday productivity tools that there’s no sense in waiting to leverage these small improvements.
In every area of operations, chat AI can be used as a tool to complement work people are already doing. In news organizations, this is just as true on the business side as on the reporting side.
A great place to start is to look for friction in every role in a news organization. Here’s a checklist of questions you can ask:
- What are processes harder than it seems they should be?
- Where do you have friction in a workflow?
- Which tasks are repetitive?
- Which things require going through large amounts of data or information?
- Which things feel like grunt work, or you wish you had an intern for?
- Things that require more time than judgment.
During my time as a professor of practice doing TV news innovation at Arizona State University’s Walter Cronkite School of Journalism and Mass Communication, I wrote about the importance of creating a stop-doing list. The premise was that an essential step to creating capacity for implementing new, better ideas and workflows was to rethink existing practices rather than endlessly adding new duties to everyone’s jobs. ChatGPT and its peers are tools that can be used to reimagine past workflows.
The potential use cases for these small-efficiency wins are vast. Joe Amditis at the Center for Cooperative Media has published a free e-handbook, ChatGPT for Local News Publishers, which features more than 50 potential use cases just for newsrooms. The list includes everything from transcribing audio/video to generating story headline options and draft social media posts, developing FAQs, summarizing long articles, and creating meeting agenda, project timelines and budgets.
On the sales/revenue side, there are just as many uses. Here’s how Chat GPT answered a prompt for ways to use it to optimize the media sales funnel:
3. Learn the ‘language’ and limitations of AI
Leveraging the potential of AI chatbots requires learning how to get the most out of the tools and also how to manage the weaknesses and limitations of this generation of AI. Two key concepts are “prompting” and “hallucinations.”
“Prompting” is the term used to describe the question, assignment or prompt given to the chatbot. The quality of the results generated by the AI is hugely influenced by the precision of the prompt. It’s the AI version of “garbage in, garbage out.” Learning how to write queries or prompts that are specific and leverage the full capabilities of ChatGPT is one of the new essential skills of using AI tools as a co-pilot in journalism, or any other job.
“Hallucinations” is another key concept, referring to the tendency of these chatbots to produce confident-sounding but incorrect answers to prompts, sometimes even including false sourcing. ChatGPT is capable of producing answers in just seconds to such a wide and deep set of questions that it’s only natural to trust the results. But there are ample instances of so-called hallucinations, so it is essential to understand that chatbots have the capacity to return authoritative-sounding but false results.
4. Ask hard questions
New technologies force us to ask hard questions. I’m reminded of the early days of organ transplants. First came the shocking medical breakthrough: the new ability to perform organ transplants. Then, given the scarcity, the cost and the risks, came all the ethical and policy discussions to create a framework for who should receive this gift of life, in what order, and based on what criteria.
With these new AI capabilities, we will have to invent new frameworks and policies that address core existing journalism concepts like:
- Authorship (Who exactly “wrote” this?)
- Copyright (Who truly owns the work?)
- Sourcing (Where did this come from?)
- Transparency (How did we create this content and what role did AI play?)
5. Develop guidelines for how, when and why you use AI
Discuss. Debate. Read. Engage. And use those conversations to create and transparently share first-draft answers to guide your teams on how, when and why you’ll use AI for work in terms of policies, practices and guidelines that you’ll co-create and regularly update.
A way to begin is to reflect on the first principles of journalism, and explore how to apply these principles to this emerging capability. What does each of these look like in an AI-assisted reporting world?
The wisest new aphorism I’ve seen is: AI won’t replace your job; but someone smartly using AI could. So let’s do what journalists have always done best: Be skeptical and be curious. Let’s invent both journalistic uses for these remarkable tools, and the guardrails to ensure we manage their risks.
Our recently published report, “Here come the machines: AI for local news” is available for free to anyone working for a Local Media Innovation Alliance subscriber company. Those not subscribed to LMIA may purchase this report for $159.
Not sure if your company subscribes to LMIA? Contact Lindsey Estes, LMA chief of staff.