You mean, a Mormon AI?
Allow me to moderate my blast. The exclamations from young men I’ve overheard around my church regarding ChatGPT focus on its supposed potential to take us forward and do new things - displacing men for positive development. But it’s evident that it will take a lot of human oversight to keep ChatGPT on track.
As a 13-years-ago me wrote in a different context when an earlier set of breathless whiz-bang assertions was coming through…
the video’s primary message came across to me as a depressing “In the near future, individuals won’t matter anymore, and the world is changing so fast that it’s hardly worth trying, because the things we do will be obsolete before we finish them.” I would say rather, “The world isn’t as different as we hoped it would be by now. We’ve tried hard, but we still aren’t there. Sometimes we’re not sure we will ever get there. The situation calls for individual men and women of courage who are determined to pull for what’s right and good – will you be one of those?”
If some man finds a way to hitch up ChatGPT and plow his fields with it like Elisha behind 12 pairs of oxen, then maybe that’s the next step in taking dominion of the earth, and I’m all for it.
Love,
2 Likes
A hot air balloon might be a good analog to ChatGPT - flashy new technology that could be greeted with claims that it would completely change the world, and that in hindsight was innovative, entertaining, a helpful tool for certain people to do certain things, but not earth shattering.
2 Likes
I suspect this is true.
Well, I don’t know what the effect of tools like ChatGPT will be, but I suspect it will impact all of our lives at least moderately. Personal computers had relatively inauspicious beginnings (I’m told by my parents that a common quip was, “What are you going to do, keep your recipes on them?”) and now many people drive a personal computer to work with a personal computer in their pocket and a personal computer on their wrist, all of which talk to each other wirelessly.
Time will tell. I’d recommend that any doubters go interact with ChatGPT themselves. Daniel, I infer from your blog that you are a programmer. Go ask ChatGPT the kinds of questions that you’d ask a fellow programmer if you were interviewing him for a job. The results might surprise you.
Again, time will tell. But if you haven’t interacted with ChatGPT, I wouldn’t recommend downplaying it. It wasn’t so long ago that SQL was the closest we had to natural language processing. Then we got Wolfram Alpha, whose NLP stunk. Then came Siri, which was pretty fair, but had a very narrow range, just kicking most things over to Google. Now we have ChatGPT, which is sort of like Siri for everything. It’s likely that the rate of improvement on NLP is going to level off soon, but practical applications are going to lag.
And I note that balloons have been in the news lately, so maybe that technology has been more influential than it might seem.
“It is hard to make predictions, especially about the future.” - Yogi Berra
2 Likes
People are already finding it very useful in a number of ways.
For example, Vanderbilt’s Peabody College recently apologized for sending an email about a school shooting that was written (at least in part) by ChatGPT. Why would they use ChatGPT for that? Because it can be a lot easier to let ChatGPT produce the content you want than to actually produce it yourself.
Similarly, I do keep seeing programmers talk about using it to help speed them up in various tasks.
In both of these cases, the final output is going to be vetted by a human before release. Those humans may or may not accidentally miss things they should change or remove before release, of course, leaving the human responsible. But who cares if they got help ghostwriting it?
From the perspective of the classic protestant technology view: “it’s a tool, and it’s all in how you use it.” If your goal is to produce content, it could be a tool that helps you. It’s somewhat similar to copying and pasting a response you’ve given to other people already when asked the same question again.
I have my doubts about this, though. Remember the fault a lot of people found with Acts29’s was not revealing Mark’s stuff was ghostwritten.
Is it honest to say that I wrote a program or letter that I had help with? How much help? I don’t really know.
1 Like
I find it easier with ChatGPT than Google to exclude what I already know, have already tried, or can’t use/acquire from the results. I’m used to googling with lots of -“exclude this” type params, too.
Have you read The Prompt Box is a Minefield: AI Chatbots and Power of Language? The beginning section tells the history of the first chatbot, ELIZA. This relatively simple program showed how quickly people “became emotionally involved with the computer and how unequivocally they anthropomorphized it” (a quote from Joseph Weizenbaum, the computer scientist who created it). The rest of the article goes on to examine some dangers inherent to introducing AI tools that mimic our capacity for sense making in an age of increasing loneliness and isolation.
Although the results of interactions with these tools might be surprising, understanding their limitations will help mitigate the powerful effect of a machine’s ability to harness human language. A machine has no soul and was not made in the image of God. Its ability to use language is derived from the source material with which it was trained. It does not have the ability to come up with novel ideas or expressions other than by chance re-arrangement of existing texts. Of course, there’s more to it than that, and it will certainly impact the creation of works with a certain form of editorial style such as that of Vanderbilt’s email. But that impact will only amplify our desire for real meaning and connection. The fake style will become more repugnant as the ease of manufacturing it with AI becomes common knowledge.
3 Likes
It seems to me that many pundits hyping ChatGPT now also hyped MOOCs back in the day (remember those)? Predictions of how MOOCs were going to transform higher education and drive universities out of business didn’t pan out, and this failed track record makes me think predictions about the revolutionary potential of AI are not going to pan out either.
The reason MOOCs failed is that they were proffered as a solution to a false problem, which was that classroom instruction was inefficient and therefore unnecessarily expensive. But the real problem was that most college students don’t know how to learn on their own and aren’t motivated to learn on their own, which MOOCs didn’t help solve at all. So the main group of people who benefited from MOOCs were highly motivated professionals who had already graduated from college.
Similarly, ChatGPT is proffered as the solution to a false problem, which is that it is inefficient for humans to research, write, and code on their own. But there is no guarantee that a text prediction AI is going to return accurate information 100% of the time, so the human is going to need expert knowledge to vet the output from the AI, which again means that the main group of people who are going to benefit are those who have the least need for assistance.
Would it help an expert to have an AI write boilerplate code? Maybe. But the experience of computer-assisted flying of airplanes offers a cautionary tale. The technology has advanced to the point that airplanes can fly themselves in most situations, but this breeds inattention on the part of the human pilots, and there have been a couple of cases where airplanes crashed because a crisis suddenly emerged and the pilots were unable to wake up and orient themselves fast enough to the actual situation to make the correct decisions.
Airplane auto-pilot is a good analogy. In fact, GitHub’s (owned by Microsoft) auto-code producer using this sort of AI is called Copilot.
But the analogy seems to prove that it does have the ability to majorly change whole industries, and “producing content” is descriptive of a lot more jobs than “flying a plane.”
I forgot I meant to share this:
I’m always so depressed seeing the random questions and answers on Quora or other places about Christianity. I don’t see how this is any worse, though. At least it answered correctly for this first question I asked.