How will Upwork address the flood of ChatGPT proposals?

[#1]

I posted two unrelated jobs today (US only, for what's it worth) and received of a flood of identical proposals that follow the same, obviously AI-generated format. Two are below, but there were many more. I remember a forum post saying that ChatGPT proposals were verboten, but it's prevalent and done openly with impunity. I asked one why he didn't simply submit a normal proposal and he said: "I am using chatGPT to help me respond better. I am not very good at explaining details but if you won’t pick me that’s fine. Have a nice day."

That response shows me that, assuming Upwork is legitimately forbidding ChatGPT in proposals and content, there needs to be a better messaging mechanism if these guys don't know they're violating the Terms. 

As a small-time client, it's annoying to pick through ChatGPT proposals to find the real ones. But more importantly for me as a freelancer, it's giving professional clients a poor experience that's going to make them avoid the platform. 

**edited for Community Guidelines**

  

[#2]

@Jonathan. :

I thnk there is a misunderstanding regarding my complaint. My complaint is NOT "I hire a Contractor; the contractor sends me a working ChatGPT solution."

My complaint is:
1. I post a Job.

2. Freelancer submits out a long, detailed, well written proposal using the right keywords / libraries with plausible pseudocode -- that all turns out to be incorrect BS generated by ChatGPT.

3. I waste time checking this BS project proposal; whereas in a pre ChatGPT world, this freelancer would not have been able to waste my time.

The problem here is: at the pre-hiring stage, in a post ChatGPT world, it takes more time to figure out who is a competent freelancer and who is incompetent, because incompetent freelancers use ChatGPT to submit plausible sounding BS that uses all the right keywords / libraries.

[#3]

I read the ChatGPT proposal in this thread's first post and I can say it's another sample of "nice but empty words". So if I'm a client I'd skip it in an instant. It has no pseudocode or anything, I can tell I don't like it from reading the first 1-2 sentences.

Freelancers are told to hone the skills to spot scam jobs at the first glance so they can scroll through jobs faster. I don't mean to say anything bad, but I think to have a similar skill (like what I have) is one plausible solution right now. At least not until the AI is told to create the proposal in a not "nice but empty words" way. Or until Upwork can really do something about it.

[#4]

Jeremy,

How many proposals did you receive? How many are suspected ChatGPT? What countries did the ChatGPT come from?

This would provide some good information as to the scope of the problem.

[#5]

Post answers both. US only. Enough were AI-generated to be notable and warrant the post.

[#6]

Jeremy,

Maybe Upwork edited out your answers to my 3 questions because the information is not visible.

[#7]

Jeremy, only the country question (answer: USA) is answered in your original post. It is hard to quantify a percentage when the only relevant descriptor is "flood".

[#8]

**edited for Community Guidelines** If I can pull direct commonalities between disparate proposals in separate job posts, and the applicant *confirms* he used ChatGPT to generate a proposal and misrepresent himself as a freelancer, then it's a problem. Scope and scale, etc are irrelevant - if one "big client" new to Upwork with a 5-figure spend potential leaves the platform because of the issue, then it affects my bottom line, as it does yours. I'd recommend focusing on how it could affect your freelancing business rather than pedantry.

[#9]

**edited for Community Guidelines** It is common for a low percentage to be perceived as a "flood", because our brains focus on the pattern that is most alarming. That is why William requested numbers.

You have every right to be irritated by the deceitful proposals that you received. And you are correct that Pathos is an important consideration for the severity of an issue. But Pathos cannot substitute for Logos - the detached, numerical analysis that determines the extent of an issue, in addition to assisting in determination of the severity.

[#10]

Jeremy wrote:

"On Upwork, you can’t do anything that’s dishonest or meant to fool others."

Okay... But ChatGPT doesn't care about that.

re: "You can’t misrepresent yourself, your experience, skills or professional qualifications, or that of others. This includes: lying about your experience, skills or professional qualifications"

ChatGPT doesn't care about that, either.

ChatGPT does not think of itself as being obligated to follow Upwork ToS.

At an academic conference here in my city a couple weeks ago, a noted author and computer scientist asked ChatGPT to provide a description of himself. ChatGPT did a fantastic job. It provided a thorough and accurate introduction to the author, in complete, correct sentences.

The author then asked ChatGPT to provide a description of a specific book he had written. ChatGPT once again provided an eloquent, convincing description of the book.

The author then pointed out that he had never actually written such a book.

[#11]

...the freelancers are using ChatGPT to misrepresent themselves in the proposal. As I provided examples of. That's the violation. If your buddy used that ChatGPT output about his little book that he didn't write in a proposal as proof of his book writing ability, it would misrepresent his skills and be a violation.

I'm unsure if you're intentionally misrepresenting the overall concept or I'm genuinely not conveying it properly, although others seem to understand the point I'm driving at. 

[#12]

re: "I'm unsure if you're intentionally misrepresenting the overall concept or I'm genuinely not conveying it properly, although others seem to understand the point I'm driving at."

I don't have any point to make.

The sense I get is that you are advocating that Upwork should have rules governing how ChatGTP-generated content is used on the platform.

I'm not at all opposed to Upwork having rules about this.

I'm simply observing the fact that Upwork doesn't have any way of filtering out ChatGPT-generated proposals or submitted work.

Any discussion about rules should include an awareness of the enforceability of such rules.

re: "If your buddy used that ChatGPT output about his little book that he didn't write in a proposal as proof of his book writing ability, it would misrepresent his skills and be a violation."

This person was a lecturer, giving a presentation. Not somebody sending a job proposal.

[#13]

Upwork can easily end ChatGPT spam. Charge freelancers $1 / application. Ends motivation for spamming ChatGPT 'solutions'.

[#14]

With the Boost Feature many freelancers spent more than $1 per job already.

What's $1 compared to win a job from "clueless" clients? I even suspect the entire system has shifted into more freelancers focusing on boost, than delivering quality work.

[#15]

Your data is out of date. Freelancer already pays a minimum of 8 connects per request, which is 1.2$! There are requests for 12 and 16 connects, that is 1.8 and 2.4$ respectively. And these are the rates without the application boost.

[#16]

Your data too. 32 was seen. 😊

[#17]

Yes. I think too.

Many freelancers bid with proposals generated by OpenAI.
Furthermore, almost clients are difficult to check those proposals.
Upwork may need to check each proposals with AI detector.

[#18]

re: "Upwork may need to check each proposals with AI detector."

I don't think that is possible.

If Upwork tries to do that, it won't be accurate.

As I client, I already have a strategy that works: I don't read proposals.

I know that many clients are accustomed to reading proposals. The truth is, right now, if clients read proposals, they're probably going to encounter a lot of AI-generated proposals. They're going to need to adjust to this reality one way or another.

Doing video interviews with candidates may be another technique clients can use to weed out incompetent candidates who sent AI-generated proposals. But that is a time-consuming thing to do.

[#19]

Of course, I am not sure that AI detector can check all proposals correctly and it is right solution.

Actually, AI is fantastic technology and I am very familiar with OpenAI and AI.

However, this problem is a lot of bogus suggestions generated by AI.
Because AI is so powerful, it turns beginners into experts in proposals.

[#20]

Hi everyone, 

A few posts have been removed for violating the Upwork Community Guidelines. While we'll continue to allow criticism, posts that come without constructive feedback or are disparaging of other members won't be allowed. Forums, like the Community, are at their best when participants treat each other with respect and courtesy. Please consider this in your future replies

As for the concerns on ChatGPT, we will continue to share these with the team. We will return and share more information with the Community if we have more information. 

We appreciate your participation.

[#21]

Definitely it will require more scrutiny from clients to ensure that they are hiring right person. However, there are tools in the market cropping up which can flag the content generated through conversational AI / CHATGPT..

[#22]

Buddy,  Proposal is one great way for them to lie, so I treat it as garbage. 

[#23]

My profile is "top rated" and 100% success rate with great reviews, I feel that since the introduction of the ChatGPT integration, many lower quality developers who likely cannot write a decent paragraph in English or address the problem the client has will either flood/overwhelm the client or be selected without merit/fraudulently.

Our investment into Upwork is being used to increase competition/race to the bottom in a way where nobody will be able to pay for the membership and make the money back. I saved up connects for 3 months and I had around 300, I used them to apply for proposals recently and I had around 40% views and no messages back. This feels very different to even a few months ago where if I sent 5 proposals I'd get at least 2-3 messages and a 60-80% view rate.

At that time I was very happy with Upwork and even upgraded membership - at the moment, I feel like leaving the platform altogether and not wasting my time. Very concerning.

[#24]

It is going to be interesting to see if clients get tired of having a gap between the cover letter and the actual freelancer that shows up for work.

[#25]

If the client gets tired? If 300 connects and spending hours applying for
these projects is a complete waste of time even when my profile is clearly
marked as 100% success rate and top rated and it's $25/hr - $15/hr has the
same issue too... why would I not get tired of that as the worker? The
connects increase per proposal, the price of connects increases, the cut is
huge... I'm not trying to be completely pessimistic about Upwork, more of a
deep concern that I cannot invest my time here. I hope it's different for
you, good luck to you.

[#26]

I agree that is not fair for many of the skilled freelancers that lose jobs to freelancers that can partically fake it. Hopefully clients can see through this.