It’s been almost a year since OpenAI, the San-Francisco lab co-founded by Elon Musk, released Generative Pre-trained Transformer 3, the language model that can produce astoundingly coherent text with minimal human prompting — enough time to draw some conclusions on whether its brute-force approach to artificial intelligence can in time allow most writing to be delegated to machines. In my current job at Bloomberg News Automation, I’m in the business of such delegation, and I have my doubts that the trail blazed by GPT-3 leads in the right direction. [Source]
Using GPT-3 for Journalism vs Literature… Some good thoughts.
Not to get too far off topic, but I’m reminded of something my Journalism Ethics professor said once that stuck with me. Basically, after Watergate, J-schools saw a surge in bright young people wanting to become “famous journalists…” Colleges began pumping out people with things other than the community’s best interest at heart. There were still a lot of good journalists even around this last turn of the century, but I had missed the “golden years” of US media (newspapers)…
I think that the internet (and citizen journalism - not to give myself away here) have accelerated this shift from having the community’s best interest at heart (or being a curious individual) toward a more “celebrity” based journalism - ala Tucker Carlson or Chris Mathews. (Perhaps bad examples, but I don’t watch very much cable news anymore. Much better ways to grok the state of the world on a daily basis…)
That the media industry is scrambling with automation is to too little too late likely. Facebook and Google ate up most of the ad revenue (mostly because they could show actual results…) I’m reminded of another time at the newspaper’s HQ in McLean, VA. (No, I didn’t touch the blue ball. But I was tempted!) I stood up during a corporate talk with a question and asked what they thought of Google Ads. They laughed and said their ads were bigger and interactive and flashy and… they didn’t get it.
On the bright side, I could see an app like Trig from @brian being useful to take a bunch of news headlines and filter out the really important stuff - or simply tag it really well. The world still needs (human) journalists - at least for the time being. Aggregating and summarizing the vast streams of data can be useful when done well by automation, though, especially in use-cases like @m-a.schenk mentioned.
Who knows. The future might be news-gathering drones and bots that constantly scour the meat-world for the precious data the algos in the background crave. Having watched the death of the American newspaper from the inside, I hope something’s figured out. Journalism is so important to a free society. The current media landscape scares and saddens me most of the time.
AP’s been publishing automated articles on quarterly earnings reports for almost a decade, and before that there was a lot of generic mad libs template stuff.
AP’s been publishing automated articles on quarterly earnings reports for almost a decade, and before that there was a lot of generic mad libs template stuff.
Yeah, sports stories (scores) were one of the first areas. I remember that back in 2007 or so?