🐈
» Forums » Freelancers » Re: PSA: Beware AI-written forum replies
Page options
25005175
Community Member

PSA: Beware AI-written forum replies

Examples aren't allowed, per the Community Guidelines, so you, reader, will need to take my word on this.

 

As in the rest of the internet, this forum is steadily filling with responses written by AI tools.

 

So, in order to "engage" in the Community, some users are using AI tools to write responses to topics and replies throughout the forum. Often, these responses include bullet points or numbered lists. They are also written in a sophisticated, matter-of-fact style, sometimes with a touch of empathy, that lends them perceived credence. To the less knowledgeable (this is not an insult, because until we learn something, we are uneducated on the topic, which is why we ask questions), these responses can seem authoritative and even logical.

 

But they are not! Many of the claims written by these tools are easily disproved by:

  • reading the Support articles,
  • reading the Terms of Service,
  • reviewing the content in the Upwork Academy, and
  • even just using the Upwork user interface.

 

For regularly active members, these AI responses can be identified at a glance, especially when they are blatantly incorrect. For new users or only occasional users, it can be difficult or even impossible to identify them. But they exist, so user beware. Honestly, despite many long-time users dislike of them, badges such as "Soaring Solutions" can be useful to gauge whether the person who posted a reply is a trustworthy source. But don't rely on the "Upgathering Votes" badges; fake news and wrong answers can acquire a lot of positive attention.

 

 

ETA: spelling

83 REPLIES 83
fe9b8d82
Community Member

LOL there's a couple of regular, prolific posters on these forums who I'm positive are doing this.

Yes, there's one right here in this thread, posting repeatedly.

And amazingly, they do not seem to realize they were caught with their first post. Such "wisdom" from someone with no profile, no experience, and the inability to write a sentence without a program.

I think they're just trolling; best not to engage.

sajal36
Community Member

So people started using it as a Co-pilot in community interaction. I believe the prime issue is bias and influence which get build up.

Don't want any ChatGPT "co-pilots".

 

I would like to see anyone using ChatGPT to generate Forum content permanently banned from the Forum.

Agree!! 

Edit: nevermind, not worth engaging


Sajal S wrote:

So people started using it as a Co-pilot in community interaction. I believe the prime issue is bias and influence which get build up.


It's not just a "co-pilot," it's putting the forum on auto-pilot; people are copying questions and posting the responses here, without doing any filtering or contributing any of their own thoughts. And for what? If you don't know the answer to a question and/or you have nothing to say, why post at all? 

Christine makes some excellent points here.

 

The thing about this Community Forum:

It is very active. There is a large number of Forum regulars who participate and who offer high-quality information tailored specifically to the needs of users who post questions.

 

That's an important factor in this whole question:

If nobody was answering questions and nobody was getting help, then looking to ChatGPT might be considered a "better than nothing" approach.

 

But that's not the case. The more apt description is "if it's not broke, don't fix it."
The Community Forum works. It is not "broke."

Can it be improved? Yes, definitely.

 

Does the introduction of ChatGPT responses from random users actually improve the Forum for people using it and trying to learn and get help?

RE:Does the introduction of ChatGPT responses from random users actually improve the Forum for people using it and trying to learn and get help?

I don't think it does. For each answer we are giving, we will need to give two. One answer to the person giving the bad info and one right answer to the person who accepted the bad answer as a solution.

sofia2008
Community Member

Many of them just started on Upwork, have no experience of the platform, have no question and suddenly know everything about anything, providing vague info, not always accurate.

Sophie,

 

It's not uncommon to encounter freelancers on Upwork who are new to the platform and lack experience. In some cases, these freelancers may overstate their knowledge or expertise in order to secure work, which can lead to inaccurate or incomplete information being provided. I think the way to prevent this is through a strict moderation control.

I find it interesting you/the program speak of new freelancers with no experience giving advice. You are new, have no jobs, and no Upwork experience.

Jeanne,

 

I'm not a program, I'm a human being. What makes you think that? It's also interesting that you think that I need to be "old" in upwork, I need to have a "job" in upwork and I need to be "experienced" in upwork to have knowledge.

sajal36
Community Member

I believe you must appreciate the feedback you are getting from members. It is the most difficult to get right advice, so you must take it as constructive feedback and correct your approach.

"I'm not a program, I'm a human being. What makes you think that?"

 

Oh, you are real.

 

Yes, you need experience before setting yourself up as an expert on Upwork. You don't have it. Now, why don't you answer my question?

Real enough to pay for the Availability badge, it seems.


Utku S wrote:

In some cases, these freelancers may overstate their knowledge or expertise in order to secure work, which can lead to inaccurate or incomplete information being provided.


They won't impress anyone here. Work is secured by actual kmowledge and expertise, not overstatements.

Sophie,

 

I completely agree with your statement. Overstating one's knowledge or expertise may help in securing work in the short term, but it will eventually backfire in the long run. Providing inaccurate or incomplete information can harm the reputation of the freelancer and can also result in dissatisfied clients.

kelly_e
Community Member

Wow. Whenever I see things that fall in the uncanny valley of writing here, I assume ESL. Hadn't occurred to me that some might be AI.

 

I'm genuinely confused. What is the *point* of doing this? I mean, what does anyone get by trying to be a Big Gun on a help forum? Where's the return on investment?

Training. It is safe, noone banned.

celgins
Community Member

I think some of it is personality-based. Some people need to appear intelligent or feel like they are an authority figure. Then, there are those who are not comfortable with their English-speaking skills or their writing, and they don't want to sound illiterate.

 

For anyone using AI-generated tools to write Forum content: Please know that the goal of these Forums--in my opinion--is to share, learn, and help others. No matter how unstructured your writing may be, or how terrible you think it is, people will try to help you. But you need to be yourself.

8467c28e
Community Member

Clark,

 

I agree with your points. Personality traits, such as the need to appear intelligent or authoritative, can influence the way people communicate online, including in forums. Additionally, language proficiency can also affect one's confidence in writing or speaking, and may lead to a reluctance to participate in online discussions. However, as you mentioned, the primary goal of online forums is to facilitate knowledge-sharing and learning among members. It's important to remember that everyone has unique experiences and perspectives to share, regardless of their language proficiency or writing style. Being authentic and genuine in your communication can help create a welcoming and inclusive environment for everyone.

yofazza
Community Member

What is the *point*

Personal satisfaction I guess 😁

 

I was tempted to do that when it was new, testing to see if it can do better than the notorious support bot, but quickly changed my mind when I realize it often provide wrong although sounds-convincing answers.

 

The long, formal, thorough, and grammatically correct text it produce also starts to becoming boring after some time. Let the moderators be the only ones to speak like that.😅

8467c28e
Community Member

Kelly,

 

It's possible that some people may create AI chatbots to participate in online forums to gather data for research purposes, to automate customer service or to provide assistance to users. However, it's important to note that not all users on help forums are AI chatbots, and many are real people who genuinely want to help others. As for the motivation behind some users who try to be seen as a "Big Gun" on help forums, it can vary. Some people may enjoy sharing their knowledge and helping others, while others may do it to boost their ego or to gain recognition within the community. Additionally, some individuals may be motivated by the desire to build a personal brand or to market their products or services.

celgins
Community Member

Glad to see this post. Some AI-written posts are glaringly obvious, while others seem to mask their AI-generated content by weaving in their own words, which often makes the entire post sound even worse.

bilal1983
Community Member

So far it is a relatively minor annoyance, with these AI posts. Thank you nonetheless for spreading awareness as not all of us knew this was happening.

 

I haven't looked too deeply into ChatGPT, and heard that this is just the beginning for it and AI in general. What else would they be taking over? More specifically, are our livelihoods as freelancers in jeopardy? How can we ensure we stay relevant?

I haven't looked too deeply into ChatGPT, and heard that this is just the beginning for it and AI in general. What else would they be taking over? More specifically, are our livelihoods as freelancers in jeopardy? How can we ensure we stay relevant?

Good questions. The same questions are being formulated in every industry and sector all around the world.


The existing and approaching generative AI storm will affect some industries more than others, but will likely impact most things in some capacity. For generative AI chatbots and the like, OpenAI has already introduced ChatGPT-4, which is apparently a lot smarter than the original versions.

 

AI/ML developers will continue to evolve their software applications and their use, so the world must evolve with it.

Bilal,

 

In my opinion the thing to remember about AI is that it can only re-synthesize what has gone before. In clever ways, maybe, but it can't replace original thinking—and if you ever come across the need to convince a client of your value versus a bot, that's where it lies—not in phrases that have never been laid down in that word order, which a bot can be trained to do, but in thinking that has never happened before you, today, thought it.

Good point, Kelly. In creative writing sectors, I see humans holding their own against AI. In other areas where computing powers, and automation are taking over, that's where the brunt would be felt.

For example in the Virtual Assistance domain, where there are admin tasks, customer service, ticketing systems, data entry, data analysis and quality assurance, very soon AI might become a more cost-efficient and quicker solution for many companies? Since I am from this niche, I know that a human-touch is what some clients prefer, but maybe as the tech improves, they might change their minds.

 

EDIT: Around 6 years ago, there was a multiplayer game (millions of players around the world, and it had a yearly tournamernt worth millions of dollars) that I played at the time (I was not a professional). OpenAI had just recently been programmed to play it. At the time, not a single pro player was able to defeat OpenAI.

 

It had the ability to make decisions way quicker than humans, and was always a one / two / three steps ahead. In fact, within the few weeks it played against humans, it kept getting better.

 

That was 6 years ago.

 

8467c28e
Community Member

Jonathan,

 

I wanted to give a quick heads up about something that's been going on in the forum lately. Have you noticed that some of the responses seem a bit... robotic? Well, that's because they are. There are a few users who are using AI tools to generate responses to forum topics. Now, I'm not saying that AI-generated responses are all bad. In fact, some of them can be quite informative and helpful. But the thing is, they're not always accurate. And if you're not familiar with the topic at hand, it can be hard to tell the difference between a well-written AI response and one that's actually written by a human who knows what they're talking about. That's why I wanted to remind everyone to take what they read in the forum with a grain of salt. If something seems too good (or too bad) to be true, it's always a good idea to do some fact-checking before taking action. Of course, there are plenty of resources on Upwork that can help you verify information, like support articles and the Upwork Academy. And while user ratings and badges can be a helpful indicator of a user's reliability, they're not foolproof. So, always be skeptical, and remember to verify before you act.

Hmmmm, Utku. Do you have anything to say that is neither a re-phrasing of my original post nor biased in favor of AI content? Anything constructive to add?

Jonathan,

 

I can say that while AI-generated content can be useful in some cases, it's important to be cautious and verify the information before taking action, especially if you're not familiar with the topic at hand. It's always best to seek information from multiple sources and consult with experts in the field, if possible. Additionally, it's essential to remain critical and skeptical of any information you come across, whether it's generated by AI or written by a human, to ensure that you're making well-informed decisions.

StanG
Community Manager
Community Manager

This is an interesting topic. For the regulars around here in Community, you no doubt witness how our moderators (or even some community members) will often provide the same reply repeatedly. Like many support teams, we have a library of answers ready to go for common questions. Something we have played around with (not in production) is using ChatGPT to make each of these responses unique. It's very narrow in that we've been testing it to use the appropriate tone and voice of a forum moderator, and not limiting ourselves to the ChatGPT interface a lot of end users are using. The idea being our moderators would still do edits before anything would be posted.

 

Now in this type of situation, is this considered a positive or a negative? The idea, as crazy as it may sound, is using ChatGPT to actually make things seem a bit more human, even though today it is 100% human driven. Yet, we're copy/pasting the same answer today, so we sound more like robots, no?

sofia2008
Community Member

For me, the issue is more with the content produced by «newly self-proclamed know-it-all» who can provide lists of tips or reasons, which are a variation of the already available ressources, but with a twist of info coming from «I don't know where» or false info. As a Forum contributor, I'd rather give info to someone than keeping track of that bad info circulating in such a way that new members could easily think that it is a formal and true answer.

 

I don't have an issue with moderators repeating the same thing over and over again (if it answers the question), as the same questions are asked over and over again.

My problem is not with anything that Upwork Forum moderators are doing.

 

What I don't like seeing is ChatGTP content in the Forum from non-moderators. i don't think the quality of what they have been posting is good.

 

I'm less concerned about the Upwork-related postings than the completely random stuff - not related to this site at all - that has been posted. Fortunately, those threads seem to be removed when reported.

Preston,

 

It's important to note that AI-generated content is still in its early stages and may not always meet the same level of quality standards as human-generated content. It's important to maintain a healthy skepticism and critical outlook when it comes to AI-generated content. 


Utku S wrote:

Preston,

 

It's important to note that AI-generated content is still in its early stages and may not always meet the same level of quality standards as human-generated content. It's important to maintain a healthy skepticism and critical outlook when it comes to AI-generated content. 


Thank you for posting such a good example of an irrelevant, AI-generated response. 

Latest Articles
Featured Topics
Learning Paths