There was an Ars Technica article about ChatGPT. I commented that ChatGPT was going to put BS artists out of work, for example, marketing people. The comment was massively downvoted, and one person replied that I obviously know nothing about marketing. Assumptions, assumptions...
My wife ran a whisky business, specializing in single cask whiskies, for more than 20 years. I helped her with a lot of marketing texts. So I asked ChatGPT: "Can you write a marketing text to sell single-cask whisky?" Here is it's response:
"Indulge in the ultimate whisky-lover's experience with our single-cask offering. Sourced from the finest distilleries and aged to perfection in a single oak cask, this rare and exclusive whisky boasts a unique flavor profile that cannot be replicated.
Each bottle is a one-of-a-kind masterpiece, with its own distinct character, aroma and taste. The single-cask aging process allows the whisky to take on the nuances of the wood and the environment, creating a depth and complexity that is unmatched by blended or mass-produced whiskies.
Whether you're a collector, a connoisseur, or simply looking for a special gift, our single-cask whisky is the epitome of luxury and exclusivity. Don't settle for anything less than the best, try our single-cask whisky today."
Wow. That is better than you would get from almost any marketing agency. ChatGPT is still only a prototype, but it is the harbinger of change. Marketing, non-investigative journalism, pop-sci articles - it is already capable of generating all of those. Moreover, it writes better than most people, eliminated the need for proofreading to catch grammatical and typographical errors.
If you get into more serious topics, then it does need some help. Here's an example: "How is uranium created?" ChatGPT replied:
Uranium is created through a process called nuclear fission. Nuclear fission is a process in which the nucleus of an atom is split into two or more smaller nuclei, releasing a large amount of energy in the process. This energy is typically released in the form of heat, light, and radiation, and can be harnessed to generate electricity. The process of nuclear fission occurs naturally in certain types of radioactive materials, such as uranium and plutonium. It can also be induced in a controlled manner in nuclear reactors.
This is wrong, but probably not a lot of people know the real answer. However, ChatGPT is able to correct it's own mistakes. I told it: "That's not correct. Nuclear fission splits uranium into smaller atoms. But how is uranium created?" It replied:
I apologize for the confusion, you are correct that nuclear fission splits uranium atoms rather than creates them. Uranium is created through a process called nuclear fusion in stars, specifically through the r-process in Supernovae. Uranium is a heavy element and is not formed through the fusion of lighter elements in the way that the light elements are created in stars. The process that creates heavy elements like Uranium is called nucleosynthesis.
And that is the correct answer. So ChatGPT can handle even more difficult and advanced topics, with a bit of supervision.
(Score: 0) by Anonymous Coward on Tuesday January 17, @10:55AM
For some reason, I was expecting to see an answer about mining and refining Uranium. But in the absolute sense of "created" your second pass seems like a good answer.
(Score: 3, Insightful) by fliptop on Tuesday January 17, @12:31PM (7 children)
Not sure I'll lose any sleep if that happens, or if lawyers get the boot too [acm.org]. For now any of the trades (carpenters, plumbers, electricians, mechanics) seem safe, but if AI is combined w/ a robot that's specialized in those tasks...who knows?
To be oneself, and unafraid whether right or wrong, is more admirable than the easy cowardice of surrender to conformity
(Score: 3, Insightful) by sjames on Tuesday January 17, @03:15PM (6 children)
Perhaps it's not quite ready to replace lawyers, but paralegals should be worried. Next will be the most junior lawyers.
The most junior of code monkeys probably have reason to be concerned.
Of course, if we replace all the junior positions, it becomes a bit of a question how we'll mint new seniors as the existing ones age out...
(Score: 2) by DannyB on Tuesday January 17, @04:55PM (5 children)
What about judges? Why can't ChatGPT answer all of the various filings by lawyers, accept the legal briefs. With some adaptation ChatGPT could even handle oral briefs. Then issue results on various motions, and even the final bench trial.
Whether ChatGPT could handle presiding over a jury trial is an interesting question. Of course, it could be a jury of your peer ChatGPT friends.
We could even start to have "instant trials" that could seriously unclog the legal system.
How often should I have my memory checked? I used to know but...
(Score: 2) by sjames on Tuesday January 17, @06:13PM (4 children)
Judges may use it to take care of the scut work, but I don't think we should be giving legal weight to the output of an AI just yet...
(Score: 3, Insightful) by DannyB on Tuesday January 17, @09:27PM (3 children)
<no-sarcasm>
Having instant trials using AI would eliminate justice. A new dystopian idea.
</no-sarcasm>
Next dystopian idea: Hello, I am your court appointed virtual public defender!
How often should I have my memory checked? I used to know but...
(Score: 3, Insightful) by sjames on Tuesday January 17, @10:11PM (2 children)
I'm not sure which is worse for justice, an AI public defender or a human one carrying over 200 cases...
(Score: 2) by turgid on Saturday January 21, @06:26PM (1 child)
We all know that computer programs are effectively free (as in cost) to run (electricity is still fairly cheap) so maybe this could help us relatively poor people have access to better justice? If I was ever on the wrong end of the law, I wouldn't want to have to sell my house to defend myself.
I refuse to engage in a battle of wits with an unarmed opponent [wikipedia.org].
(Score: 3, Insightful) by sjames on Saturday January 21, @09:44PM
There is that too. Justice you can't afford is no justice at all. If you have to mortgage your house to get it, you have already been punished before you even get to court and without regard for the verdict.
(Score: -1, Troll) by Anonymous Coward on Tuesday January 17, @02:27PM (4 children)
It's just a two minutes hate circlejerk about Musky, AI, Republicans, etc.
(Score: 2, Funny) by Anonymous Coward on Tuesday January 17, @02:55PM (3 children)
(Score: 2) by Freeman on Thursday January 19, @08:56PM (2 children)
I mean, have you looked at the comments section on Ars Technica? It's very one sided. Yes, the person you replied to is very biased. Doesn't make Ars Technica, especially the commenters, any less biased. Still, I like the topics they cover and for the most part the authors have brain cells they make sparks with.
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 1, Funny) by Anonymous Coward on Friday January 20, @07:54AM (1 child)
Ferkin Christians! Thank god they believe in Hell.
(Score: 0) by Anonymous Coward on Saturday January 21, @06:23PM
It doesn't seem to help though.
(Score: 2) by Freeman on Thursday January 19, @08:59PM (1 child)
All the same time ChatGPT is interesting, scary, and kind of just whatever. Actual AI has a very long way to go to be anything like people fantasize about. Unless you fantasize about an AI malfunctioning and doing very bad scary things. Which, I mean is probably just the AI doing what it's been programmed to do. Just that the programmers sucked.
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 3, Interesting) by turgid on Saturday January 21, @06:28PM
I reckon it would be pretty trivial to set such a thing off into some sort of positive feedback loop that would cause unimaginable mayhem. Luckily I'm not the sort of person who would actually do such a thing. Someone else might, though.
I refuse to engage in a battle of wits with an unarmed opponent [wikipedia.org].