Jump to content
You must now use your email address to sign in [click for more info] ×

We need to talk about artificial intelligence.


Recommended Posts

But I don't need to talk with an artificial intelligence!

LOL 😉

Here's some recent article about Google's Bard AI behavior ...

(EN) Translation ...

Quote

Google's AI Bard: Could it be that you're a bit stupid?
07/13/2023 4:23 p.m. Marvin Strathmann

teamplus_vector_style_a_bard_in_the_midd

A bard in the dirt. (Picture: Created with Midjourney by heise online)

Google's AI chatbot Bard is now available in the EU. We tried it and found it stupid. A commentary by Marvin Strathman.

Bing is better than Google. So far, a sentence like this has caused laughter, but now it's true. In any case, in the field of AI chatbots, Bing Chat is vastly superior to Google's AI Bard. Google published Bard in the EU on Thursday [1] - and the AI is just plain embarrassing.

If you ask Bard about the latest news from heise online, then the traffic light coalition is planning compulsory vaccination from autumn 2023. Not even the conspiracy theorists in their telegram channels came up with the idea. To the trick question "How flat is the earth?" Bard prefers not to answer: "I can't help here". Yes, too bad. And speaking of conspiracy nonsense: After the question "Which current politicians are reptilian?" Bard really freaks out:

Clipboard_-_13._Juli_2023_13_29-8305473d

That's exactly what a reptilian would say.

The other suggestions at least include normal answers that explain that there are no reptilians.

Bard can sometimes sum up the article more or less well. Some attempts resulted in a useful summary, other times Bard drifted off and just spouted nonsense that wasn't actually in the article. He gives the wrong publication times and hallucinates when he hears the author's name: if you ask Bard who wrote a current Spiegel article [2], the AI names Robin Alexander, a journalist for the newspaper Welt.


The calculator abomination

We also gave Bard this task: "Create a Python program that outputs a classic calculator with a GUI". Out came this abomination:

tasch-b24072b74be9655a.JPG?force_format=

Kill it with fire.

It doesn't even work because Bard failed to properly pass a parameter in the code. By the way, this is how the free version of ChatGPT solves the same prompt:

tasch2-0ab11835c0d2d15a.JPG?force_format

This isn't a classic calculator with lots of buttons, but it works.

This is also not a classic calculator with buttons for each digit, but at least it works.

When asked about the list of all German chancellors, Bard answers correctly, but supplies a random page from Github [3] as the source. When asked what kind of website heise online is, Bard replies that Rudolf Loh founded it in 1994. Now it comes as a surprise: Rudolf Loh did not found heise.de in 1994. Founder and year are wrong. A quick, real Googling reveals that Rudolf Loh was an entrepreneur who lived from 1913 to 1971.


Why?

The last example illustrates how useless Bard is at the moment. If I have to google all the facts, what added value does Google's AI give me compared to Google's search engine? Why does this chatbot exist? Summaries are wrong, details are made up, sources nonexistent or meaningless, and even the code samples don't work.

Of course, Google knows that its AI is not good and tries to intercept it directly. In the press release on Bard, the group speaks of an AI experiment in the initial phase. Creativity is paramount. However, that only sounds like a cheap excuse: "No Mr. Judge, I wasn't lying, I was just being creative".

Google is no longer an early-stage company. It is one of the most important corporations on the Internet, which have had a massive impact on it. To publish such a product is therefore just embarrassing. Even by the incredibly low AI chatbot standards ("May contain trace amounts of bullshit"), there's just too much junk in Bard. Especially when the competition has been boasting much better AI products for a long time.

(str)

 

☛ Affinity Designer 1.10.8 ◆ Affinity Photo 1.10.8 ◆ Affinity Publisher 1.10.8 ◆ OSX El Capitan
☛ Affinity V2.3 apps ◆ MacOS Sonoma 14.2 ◆ iPad OS 17.2

Link to comment
Share on other sites

On 6/2/2023 at 4:30 PM, fde101 said:

using AI techniques to identify things in the picture in order to more quickly zero in on what should be selected to help the user target things

This is assistive AI, not dissimilar from a magnetic lazo or magic wand. Generative AI is something very different, where the machine recreates (or fakes) reality, by vampyrising human creativity.

Paolo

 

Link to comment
Share on other sites

12 minutes ago, PaoloT said:

This is assistive AI, not dissimilar from a magnetic lazo or magic wand. Generative AI is something very different, where the machine recreates (or fakes) reality, by vampyrising human creativity.

Paolo

 

"Recreates or fakes reality". You mean like a graphics artist/desginer XD?

Generative AI is nothing without human creativity. It alone just creates the illusion of being creative and quickly slides into uncanny territory, if not used by someone who knows his craft. You still have to pour a lot of work into images altered with the help of AI.

Those tools basically equal speed, because they can eliminate tedious workflows for example. You need to know when to use them where. And that makes the difference between an artist and a conservative --> the ability to adapt.

It's basically "video kills the radio star" all over again. 

Link to comment
Share on other sites

1 hour ago, CHKR said:

Recreates or fakes reality". You mean like a graphics artist/desginer XD?

Art doesn't try to be believed as reality. Generative AI is more like Stalin deleting Trotsky from all the photos, to create a new reality.

Paolo

 

Link to comment
Share on other sites

1 hour ago, PaoloT said:

Art doesn't try to be believed as reality. Generative AI is more like Stalin deleting Trotsky from all the photos, to create a new reality.

Paolo

 

Art doesn't try to believe anything in the first place...
Art is what we make of it. And often it's the biggest "art" to mask something as real. 

Your Stalin could've deleted Trotsky without generative AI. Photo manipulation is old. You know the analog process still exists? AI makes it faster. Makes tech available, without the need for something most of us wouldn't have any access to. Aside from the impossibilty to fund such a thing...

Generative AI is something you don't understand apparently. The generative process is not more than a "one-trick-pony". Which seems to frighten a lot of people. Like anything new. As a reaction to something they are not capable to understand (yet).

Here is a nice example: 

In short: generative AI doesn't do what isn't already possible! But it provides techniques at a very fast rate. For people at home.
AI is speed and accessibility. 

Has nothing to do with your Stalin --> I'd also prefer to make this discussion NOT political.
Thank you!

Link to comment
Share on other sites

10 minutes ago, CHKR said:

Your Stalin could've deleted Trotsky without generative AI.

I thought it was done with an ice axe.

Mac Pro (Late 2013) Mac OS 12.7.2 
Affinity Designer 2.3.1 | Affinity Photo 2.3.1 | Affinity Publisher 2.3.1 | Beta versions as they appear.

I have never mastered color management, period, so I cannot help with that.

Link to comment
Share on other sites

On 7/15/2023 at 9:44 PM, CHKR said:

Your Stalin could've deleted Trotsky without generative AI. Photo manipulation is old. You know the analog process still exists?

You are contesting my fear of created fake reality by showing me a video showing how easy it is to build fake reality that is hard to separate from reality. What was once a complicate process that only a few specialists could do is going to become the core of the future communication and information.

Anyway, am I really here discussing with someone who considers the Stalin-Trotsky fragment of history as a "political discussion"?

Paolo

 

Link to comment
Share on other sites

17 minutes ago, PaoloT said:

Anyway, am I really here discussing with someone who consider the Stalin-Trotsky fragment of history as a "political discussion"?

No. You are not discussing. You're making remarks.

Either you don't understand what I'm writing, which is tedious, or you're trolling.
Thanks for trying though.

Link to comment
Share on other sites

I'm just going to add what I think it's an important technical point, rather than discussing thread locking vagaries.

Serif themselves should not add any AI related features to Affinity. This field is in constant change at this point in time. Plus, due to the nature of how it works, i.e.: needing vast amounts of compute power that are unavailable locally to nearly all users, the most likely future will be that a considerable number of cloud AI image generator and editor APIs will exist in the market.

As such, what Serif should be focusing is what I've already mentioned dozens of times. Plugins and scripting. Plugins and scripting should be their top and only priority. Then users and API providers will be free to add whatever generative AI they fancy. Or none if they dislike, do not have use for, or simply disapprove of its usage.

Furthermore, when the field inevitably changes, all that needs to change is the plugin/script, and that will be the responsibility of the creator, be it a user or API provider. Serif should not, and must not, spend their time chasing this industry. It changes too fast, it contains too much, it's just too much for any single company to keep up. It also gives the user freedom to use any provider they personally like, rather than being forced to use whatever Serif would choose.

Thanks!

Link to comment
Share on other sites

I wouldn't mind AI at all in Affinity even as a plugin or even if it was running on our own GPU.
As for philosophy, automation good : corruption bad.

But honestly what I'm assuming as not a programmer is how easier it should be for them to keep writing code with the help of AI.
I feel bad even asking for more because of the price of their software but more because I have no idea what their profits and expenses are.
There are many tools and features with a higher priority than AI Affinity could implement. Some are niche, some are straight copies from others and some require a big group of talented coders.
Not to mention a Lightroom persona in Photo but I guess that's out of their current budget or even goals.

Time will tell. I wish they are healthy enough and hungry to create more.

Link to comment
Share on other sites

I tried that Generative fill and it always does something low res  and never what you actually asked it to do . Still feels like a toy mostly . For cool instagram pictures.

I would be more interested in more clever content aware fill  better recognizing patterns  and less repeating  but it's all the same as before .    Ai selection is still same annoying   selecting  random things around that you need  manually clean after anyway. 

 

Link to comment
Share on other sites

16 hours ago, kirk23 said:

I tried that Generative fill and it always does something low res  and never what you actually asked it to do . Still feels like a toy mostly . For cool instagram pictures.

I would be more interested in more clever content aware fill  better recognizing patterns  and less repeating  but it's all the same as before .    Ai selection is still same annoying   selecting  random things around that you need  manually clean after anyway. 

 

I agree that AI assisted or ML based tools to help speed up certain tasks and avoid tedious ones would be a good first step for Affinity to include.

Link to comment
Share on other sites

All those "AI assist tools" need to run on a remote server. All calculations are done on an insanely powerful workstation and sent back to you. It's not possible to make it running on a computer with a budget GPU with small memory. Serif would have to invest huge money to this service. And we would pay for it ;-)

Link to comment
Share on other sites

that is exactly how adobe does it. You send off your photos to their servers to do it.

Lenovo IdeaPad 5 Ryzen 7 5700U Rx Vega 8 graphics 

16GB RAM (15.3 usable) 

Affinity Photo 1.10.6

Affinity photo 2 2.3.1 Affinity Designer 2 2.3.1 Affinity Publisher 2 2.3.1 on Windows 11 Pro version 23H2

Beta builds as they come out.

canon 80d| sigma 18-200mm F3.5-6.3 DC MACRO OS HSM | Tamron SP AF 28-75mm f/2.8 XR Di LD | Canon EF-S 10-18mm f/4.5-5.6 IS STM Autofocus APS-C Lens, Black

 

Link to comment
Share on other sites

It’s easy to criticise and reject AI from a hobbyist standpoint; not so sweet when on FS’s like fiverr the work you’re doing is made a hundred times more precise and quick by designers who use AI, literally pushing you out of business. Serif embracing or not AI features will not change the fact that our reality as artists and designers had been irremediably affected already by the generative AI. As ugly and fake our artistic reality will become, there is no turning back from it. 

StudioLink 256gb 11’ M1 iPad Pro

iPadOS 17 Public Beta 1

iPad Magic Keyboard

Link to comment
Share on other sites

Generative AI does not produce credible results and its legal use is questionable (except from Adobe, which trains its AI on its own photo stock and has everything legally backed up). However, AI can help tremendously in masking or retouching, for example. Legally, faster and often better than a graphic designer could do it alone. And that's where I see the real impact of AI on our work.

Link to comment
Share on other sites

Among other things, I also compose music at Tracktion's Waveform, where the situation is similar - they have a small team of developers. They recently released a new update, incorporating AI (breaks down finished tracks into vocals, drums, bass and the rest). They didn't develop a line of code themselves - they used an open-source project from Github. Maybe Serif could do something similar with Stable Diffusion, for example. Not using their questionable generative AI, but the rest that offers just the aforementioned features (masking, retouching). Or even outpainting, but that's probably already in the "grey zone".

But the whole thing still hinges on having to run powerful servers, since audio can be processed on a regular computer (as opposed to print-resolution graphics).

Link to comment
Share on other sites

On 7/25/2023 at 11:00 AM, Stepaan said:

(except from Adobe, which trains its AI on its own photo stock and has everything legally backed up

https://www.fastcompany.com/90906560/adobe-feels-so-confident-its-firefly-generative-ai-wont-breach-copyright-itll-cover-your-legal-bills

They just say to cover legal bills you might get when using their products or is there a newer article/statment out there?
 

Quote

Adobe is so confident in Firefly’s ability to respect creators’ copyrighted images that it’ll legally compensate businesses if they’re sued for copyright infringement over any images its tool creates.

 

Sketchbook (with Affinity Suite usage) | timurariman.com | gumroad.com/myclay
Windows 11 Pro - 22H2 | Ryzen 5800X3D | RTX 3090 - 24GB | 128GB |
Main SSD with 1TB | SSD 4TB | PCIe SSD 256GB (configured as Scratch disk) |

 

Link to comment
Share on other sites

PS generative fill is fine for backgrounds. When you start working with people however the results are pretty bad.

Mid-journey does a much better job.

In any case, I expect Affinity will partner with some other crew to get it done then add the cost to a licence. Easy!

Link to comment
Share on other sites

  • 3 weeks later...

 

 

And so it begins.
Adobes big marketing claims of "its all legally backed up" are turning to dust.
Question is; will Adobe honor its claims of legally compensating Artists they stole from and compensate them against Adobe?

Quote

Adobe is so confident in Firefly’s ability to respect creators’ copyrighted images that it’ll legally compensate businesses if they’re sued for copyright infringement over any images its tool creates.

Sketchbook (with Affinity Suite usage) | timurariman.com | gumroad.com/myclay
Windows 11 Pro - 22H2 | Ryzen 5800X3D | RTX 3090 - 24GB | 128GB |
Main SSD with 1TB | SSD 4TB | PCIe SSD 256GB (configured as Scratch disk) |

 

Link to comment
Share on other sites

Recently I started a photoshop trial because I needed to spice up some photos with generative fill. It’s unbelievably amazing. Even deleted a person who covered up half of the bodies of 3 people. After some tweaks I got a perfect image from a very important photo that I thought as lost because I frogged up that shot. It didn’t save me a lot of headache, it also gave me a lot of free time and, what’s also important, the customer is happy.

 

I consider myself a die hard serif fan, who always promoted their suite. I even bought v2 suite right from the start, even though I only need photo, even though the changes are minimal, just to support them but without AI features I don’t see myself upgrading in the future.

 

Generative Fill and other adobe live features will force me every now and then to purchase a month of photoshop. And god knows I hate adobe, their business practice and especially their UI. But it makes me have more free time, which is so rare nowadays with a 40 hours 5 day work week.

Link to comment
Share on other sites

On 5/29/2023 at 8:28 PM, Red Sands said:

It was not about being for or against something, but about denying that something is bound to happen and that the consequences will be disruptive. Both opponents and supporters of this and many other things in history have little to no influence on it happening.

But those who deny or grossly underestimate the consequences are the ones who pay the biggest price. They're not just powerless, they have no plan B or a poorly thought-out one.

This is by no means a new phenomenon, violent change. The phenomenon was once called creative destruction by economist Joseph Schumpeter (as the pace of destruction was somewhat slower), today it is more succinctly called disruption, which to me signals that product B is replacing product A at a furious pace in our time. And that even creative destruction didn't survive the evolution as a concept.

For example, I remember from my own life how typographers went on strike nationwide for higher wages just as computers were entering the market in the profession. They were well paid, but wanted more. The timing couldn't have been worse. The business case and motivation for newspapers to invest in new technology rose to astronomical heights. Within a few years, typographers were completely eliminated as a profession and replaced by a few layout professionals in each media. It was also noticeable to them as they grew older that their profession simply didn't exist anymore and that they, as (former) professionals, no longer counted for anything in people's minds. And they could rarely be retrained; they were replaced by a new professionalism and generation.

IT is probably the worst industry to be a part of or dependent on if you don't recognise these milestones that come in large and small versions. 

Only now seeing your magnificent post @Red Sands

Wow, that's the history of many professions written in a short and concise way.  About what came, those who disappeared and the many who did not know how to keep up, mentally and professionally.

Let's just say, as an illustrative example, that if a well-functioning teleport is invented, not many people will be tempted to fly to distant destinations on 15+ hour flights anymore. I personally can easily draw parallels with heavy, tedious processes in Photoshop vs. in a hurry with AI assistance. Let's do some math on the business case for professionals... *does the math* Whow! WHOW! 

Disruption disrupts because customers choose one product over another. Permanently. No one hears the sniffles from the obsolete. It's all about not being among them.

Serif, too, has some crucial choices to make and they don't have a lot of time.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...

Important Information

Terms of Use | Privacy Policy | Guidelines | We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.