All AI technologies feast on data to work well, especially generative AI (genAI) systems. GenAI has an almost insatiable appetite for ingesting source data in striving for breadth of topics that it can address as well as the natural language of prompts and responses: OpenAI’s ChatGPT3, for example, is trained on a dataset containing an astonishing 175 billion parameters and millions of documents.
Often, the source training data for genAI contains original journalistic content, but journalism can also be the generated output. The relationship is symbiotic and has the potential to reinvigorate the ailing news industry.
US Bureau of Labor data shows that the number of working US journalists has fallen by over 50% since 2004. The Washington Post estimates that an average of two local newspapers close every week, creating more so-called “news deserts” where communities lack a local news source. Media organizations are struggling to compete in the digital age, with implications for democratic processes and local economies.
But there are signs of change. The Associated Press (AP) recently announced a two-year deal with OpenAI, the parent company of ChatGPT, which is a broad news-sharing agreement between the companies. OpenAI also announced a partnership with the nonprofit American Journalism Project that will provide $5 million in funding for local news reporting, assisted by genAI.
News organizations were early adopters of AI technologies. For years, the AP and others have used AI for generating formulaic content such as sports and financial results reporting. BuzzFeed, the BBC, and many others are now experimenting with genAI to help create everything from earnings reports to local election coverage. BloombergGPT is a genAI system tuned for finance.
But the symbiosis isn’t always seen as positive. The revolution exemplified by the AP-ChatGPT deal could be replicated in other content-rich industries if IP licensing and ownership rights can be hammered out. One sign of this challenge is the myriad lawsuits already cropping up against genAI providers. Companies as diverse as Getty Images and News Corp are filing such suits, claiming IP infringement on content used for training systems.
Action For B2B Marketers
B2B marketing leaders can learn from media companies, which have shown the risks and rewards of adopting genAI for content creation. Careful experimentation is key, with an incremental approach to adoption, especially in light of pending lawsuits against genAI providers and evolving views on ownership rights for the content produced. Make sure that you have adequate governance processes in place, including who is authorized to experiment with genAI and guardrails to ensure your IP protection. Foundationally, understand the value of your own data and recognize that the value of genAI will hinge on the quality of the data used to train it.
This post was written by Forrester VP, Principal Analyst Ian Bruce and it originally appeared here.
Read the full article here
