tl;dr: Had this up for a while while working on other stuff. Iâve thought about the challenges Bubble has in terms of enabling an AI helper and I donât think they involve business logic of specific apps at all, I read the Tâs & Câs for the first time in a long time (and they continue to be quite fair and reasonable but donât really limit their ability to do internal stuff with your workflow logic if they did need to), and youâve already agreed not to sue over anything in the Tâs & Câs.
2 Likes
GPT4âs version of âhey, whatâs the tl;dr on thisâ says:
The legal definition of a copy in copyright and trademark law is highly variable by context, and itâs not necessary to include someoneâs app in a training dataset to accomplish the desired task of training a large language model like GPT-4. Whatâs missing is the blackbox part of Bubble, which is the representation of a Bubble app and its individual components. By using Bubble, you have already agreed to settle any legal claims through arbitration, and you retain ownership of your Direct User Content, including Bubble Site workflow and design. Bubble claims no ownership of your âappâ and isnât going to sell it, but it does claim some rights to analyze things, including aggregated Direct User Content.
(so I would say that my summary was pretty darn good and itâs entirely possible that I am just a large language model)
Its summary based on section 7 of the Bubble Tâs & Câs is essentially the same as mine, so Iâm just going to let GPT-4 post as me to the forum going forward.
4 Likes
@zelus_pudding letâs say you opt out of allowing Bubble to use your app for AI training purposes (assuming this becomes a thing). Bubble then introduces a way to build apps using conversational chat that would cut down app development times by weeks and days. Are you going to refuse to use that feature? If thatâs the case I think they should introduce the feature only to users who granted permission to use their apps as training material.
1 Like
Here come the silly comments!
< enable reddit mode />
1 Like
Jumping in with a bit of context and a few thoughtsâŚ
First of all, great discussion. I donât have answers to all the questions, because weâre still figuring things out ourselves: AI is changing very quickly, and thereâs a lot of product, technical, ethical, and legal questions we need to answer. We are still very much in learning mode.
The point of the AI residency weâve announced is to get more visibility into some of the product and technical questions. At this point, we have a lot of ideas on how to train AIs to be helpful co-Bubblers, but we havenât tested them. The goal of the residency is to prove out concepts, not to release production features. We wonât release anything until we have clear stances on the legal and ethical questions as well, and my working assumption is that we will most likely throw away any model that gets built during the residency, and then if we decide to launch a production feature, we will re-create the model using what we learned.
Weâre committed to implementing AI in a way thatâs net beneficial for our community and in line with our mission of empowering more people to be creators. This conversation is helpful, because understanding whatâs top-of-mind for you all with regards to AI helps us figure out how best to do that, so I appreciate all the commentary! (And Keith, your experimentation is really cool)
What Iâm hearing is thereâs a lot of excitement about AI mixed in with a healthy dose of fear. The biggest fear being, will AI devalue your hard-won expertise as Bubble developers, and create a flood of cheap competition?
This is just my opinion, and I donât have a crystal ball here, but my bet is no. Github Copilot doesnât replace the need for programmers, because it takes a programmer to understand the output and give the AI feedback. Likewise, I donât think any AI we build is going to replace the need for Bubble skill and expertise.
Our belief has been from day 1 that making software cheaper and faster to build on-net creates opportunity, because the easier it is to create software, the more projects get off the ground that otherwise wouldnât have been able to be started. Since then, weâve seen the No Code community explode⌠and during that same time period, the demand for traditional programming has gone up, not down. I would bet AI plays out the same way.
I do think it is important we keep on evolving Bubble to get easier and more powerful to use. I see âBubble gets obsolete, people stop wanting to use it for things, the community and platform dies offâ as a bigger threat to the long-term value of the investments youâve all made in the Bubble ecosystem, than âBubble gets so good at making app creation easy and robust that your expertise no longer mattersâ. So I hope that if we do get to a point where thereâs a choice about letting us use your data (which is not inevitable; I am not convinced that training on our existing set of apps is the best technical approach), that weâd get a lot of enthusiasm about helping out. The Bubble team is firmly committed to the goal of having Bubble be a long-term, stable platform, and I think part of achieving that involves investing in new technologies like AI.
Again, though, we really appreciate all the input, and feel free to keep discussing on this thread â we may or may not reply again but weâll read the posts
10 Likes
Glad you enjoyed the tech demo, @josh! Seriously the leap from GPT3.5 to 4 is monumental. With a little more attention (longer convo/prompt length), these things will be fully capable of pretty amazing transformations with just a preamble prompt because they are such excellent one-shot learners.
Also, GPT4 can actually tell you what it learned during a given conversation and express it in a condensed way. Itâs fairly shocking how little new info it needs for certain tasks.
3 Likes
So I thought todayâs webinar was pretty good. Itâs always interesting to see other folksâ projects.
Josh,
I deeply respect the team, company and product youâve built. From a technical standpoint, your team has deployed what is in my opinion a flexible business building tool without equal. When it comes to how you listen, treat and respond to your customers, I find your team to be among the best not just in the industry but overall.
Iâm not embellishing - when the community spoke out against price hikes over the last few years, you guys couldâve just said âtough luck, thatâs the price if you donât like it you can leave.â But you didnât do that, instead:
- the first time around, you allowed folks to retain cheaper grandfathered pricing and,
- this second time, youâve notified the community well in advance and are taking months to gather feedback from us (Iâve personally been reached out to twice at this point).
So I genuinely trust that you guys will do your best to do the right thing when it comes to creating and rolling out a BubbleGPT of any form.
That said, I disagree with the conclusion that Bubble developer jobs in an AI dominated future will be just as plentiful and as well paying as they are now. The two intuitions supporting that idea are flawed or just obviously wrong - those being:
- AIs are useless without expert âdriversâ, therefor existing experts will still be employed to drive those AIs.
- The explosion of No-code tech has not suppressed traditional programmer job demand so, similarly, AI wonât displace Bubble developers.
AIs do not need expert âdriversâ
To the first point: the truth is experts need not apply. Weâre not needed anymore. Itâs that simple. Exhibit A: In his video ChatGPT changed how I work in After Effects FOREVER Jake in Motion explains:
I just used a brand new tool to get chat GPT to write me three actually useful After Effects scripts⌠Prior to this video I had absolutely zero experience with scripting in After Effects so this is absolutely something that you can do even if youâve never touched scripting yourself
A minute later he cuts to a clip of another YouTuber saying:
Is this what a developer feels like?
Thereâs a million examples of this sort of âI donât need an expert anymoreâ videos circulating so trying to sell anyone on the idea that expert âAI driversâ will still be needed is outrageous if not simply uninformed. And how about Exhibit B, back in 2019 I hired @vini_brito to build a plugin for me. Now, before I hired him, several developers quoted me a price to build it including @ZeroqodeTeam. Well guess what - I donât need em anymore. Iâm sorry Vinni, I know you know I love you but @keith just demonstrated how Bubble plugin developers just became too expensive for anyone to call. Bye-bye job.
This is fun right?
Oh but I release plugins for free because the marketing and good-will blossoms into real gigs that pay. Exhibit C @keith the last time you made a thing for me (which thank you again by the way) I ended up donating like $20 or $30 bucks to you. Real talk, the amount I donate towards âkarmawareâ drops fast when itâs understood the help came from a free to use AI rather than the effort of an intellectual.
No-code is still not taken seriously. AI is.
Citing the muted effect no-code has had on software developer demand as the harbinger of how AI will impact the bubble developer job market is awkward at best. The reason that demand for traditional software developers did not diminish since you started Bubble in 2012 is because most businesses still view no-code as a toy. Iâve built entire projects, written blog posts, and have made MVPs to convince some of my traditional programmer friends just to try Bubble. They wont do it. They refuse to use Bubble because âItâs a toy and only serious businesses are built with codeâ.
And Iâm not just talking about legacy projects. Iâve been scoffed at in job interviews for suggesting the use of no-code in new projects.
I mean câmon, you guys even put out this pdf to help entrepreneurs convince their investors itâs okay to build on Bubble.
No-codeâs failure to steal traditional programmer jobs en masse is in no way representative of how AI is already threatening employment for folks everywhere. By now Iâve made my point - I donât want my applications to be used in a way that diminish my income. Even if the effect of my unwilling and uncompensated contribution is 1/1 millionth of the final model, I donât want it in there. New tech is always around the corner - I get it - but even the Luddites werenât raging against a machine trained on their creativity. This is different and frankly it would cross the line for me.
Iâm going back to working on my plugin now⌠itâs something Iâve been working on for months. Cheers to hoping GPT doesnât cheapen it 
2 Likes
Youâre welcome, @zelus_pudding. And thanks for your previous tips!
Look on the bright side: at least GPT-4 can help you with your plugin code!
1 Like
Well well, would you look at this. A petition for a moratorium to pause training of AI models more powerful than GPT-4 signed by, among others, the guy who founded OpenAI. Makes sense.
1 Like
The thing Iâm left wondering (and Iâve seen multiple examples of this) is âwhere does Bubble find these people who are (1) not ding-dongs and (2) might need help but donât show up in the forum?â
Forget AI⌠It seems thereâs a whole market thatâs inaccessible-ish to plugin devs. I guess the Silicon Alley hustlers are opaque? Like, where do they congregate? Are they just cheap? (No offense intended.)
These could be businesses/apps that have been submitted to the showcase. Also, bubble has all sorts of analytics on whoâs doing what in their apps. Also also, Eric Ries - the guy who literally wrote the book on âlean startupsâ is an investor in Bubble, so a network of y-combinator type companies is aware of and occasionally trying Bubble.
1 Like
@zelus_pudding : Iâm not one for manifestos, nor for boycotts⌠but I basically share this safety concern. Earlier GPTâs were just porn-factories. (And why limit our access to that, you Mormon scuds!? [Iâm talking to OpenAI here.))
But I have concerns around models beyond GPT-4 that might have massive ramifications. I mean, I tried to jailbreak GPT-4, and it seems completely compliant⌠but also rather spooky (âWHY IS YOUR NAME SALT AGAIN!?â). But Iâm not so sure about the next generation of these children of person-kind.
Iâm not concerned about the economic impacts of this (mostleee, because Iâm already somewhat benefiting from the force-mulitplier benefits of them), and I donât really see emergent behaviors of concern from them⌠but I see tiny traces there? ⌠but I had ZERO concerns about that in previous versions, but have SIGNIFICANT concerns about that in GPT-4 (as do GPT-4âs creators or else they wouldnât have tested that).
Itâs all just a language game until someone gets hurt, right?
Blah, blah, blah.
2 Likes
Yeah, @zelus_pudding, what a bunch of dicks, amirite? (Iâm serious RN.)
Oddly, I find myself in support of that petition. Signed.
(Let the record show that this is but a small intersection between my own concerns and those of Elon Musk.)
3 Likes
The petition is against unchecked use and human advancement of AI. Itâs a response to big tech pushing AI without thinking of the long term ramifications of AI growing out of control and, I canât believe Iâm saying this; become a version of Skynet.
The research team that tested for GPT4 for Advanced General Intelligence actually concluded that GPT4 should not be released to the general public but pressure from Microsoft forced OpenAI to do otherwise.
Anyway Iâm getting ahead of myself. What Iâm trying to say is I totally support the petition because recent history has shown how lacking governments are in actually governing advancements in tech. Case in point: Rampant fraud in the Crypto space.
This is where i diverge from you though. As someone who hustled his way through adulthood (I left formal education in my teens to support my family) i learned new skills from experience and reading, continuously trying new things. It took a long time for me to be (career wise) where my peers who were able to continue their education reached a decade earlier.
Hence i am a strong supporter and advocate of self-learning and any advancements in how information can be shared. AI can be dangerous but human mindsets are worse. That said you are bordering on fear mongering with your hyperbole.
I am emphatic to your anecdotes but you underestimate the impact of No Code. To share my own anecdote, I can tell you that I am taking my competitors by surprise at the value quality i am offering clients in my sector.
I like to believe that a majority of the people here are smart enough to adapt to the new landscape that AI will scorch in the next few months whether we like it or not.
Wouldnât it be better to steer the conversation to support the Bubble team to build appropriate guard rails while AI enhancing Bubble.
They obviously have strong connections in the tech space and with strong support from the community Bubble can lead the way to AI âgovernance and controlâ in the world of No/Low Code.
3 Likes
There are a lot of good and very reasonable debates with AI, ethics and future concerns. However my view with Bubble, and inevitably other software tools leveraging AI (even Bubble competitionâŚ), is if I can build and improve my app quicker then I am 100% all for it.
For freelancers and agencies, if you can build more apps quicker, then you should also be able to make more. From what I have seen is the demand for quality app development is always high, regardless of whether it is built on Bubble in 10hours or on RoR in 50hours. Feels like complaining that the horse plough is going to get replaced by a tractor with an engine - get more output and charge on results and value, not hours.
Plus having gone through a Bubble freelancer phase, I can confidently say that a lot of individuals struggle to properly explain the app that they want built so I canât imagine them working with a âyes sirâ type of AI and it producing quality work.
3 Likes
Havenât looked into this much but if I were in this space and had the industry leading tech I would love to force all my competition to pause their work for an extended amount of time too
1 Like
Love it! How do we do that? Seriously asking.
1 Like
Honestly i wouldnât know where to start. AI is advancing so fast because big tech is aggressive in pushing out AI services. Maybe a less zealous (stages) approach is a win win to supporters and otherwise?
What i do know is that from what @josh wrote earlier, i think Bubble is heading in the right direction if they are doing as they claim. Listening to users and seeking feedback.
Perhaps a good start for Bubble is to first push out AI to bolster app building and optimization instead of a something like a prompt to app template generator. The forum is filled with questions about optimization. An AI that can suggest optimized workflows which we can then pick and choose from before generating the workflow might work.
An AI trained in both the Bubble manual and posts from the forums will be great too. Makes searching for information way easier.
2 Likes