Does AI make us more efficient… or more dependent? Who is driving your brush?

[Amybeth M. ACI G7]

I say this out loud to myself — and honestly to anyone who will listen:

Is AI actually making us more efficient… or more dependent?

I’ve been working with partner models like Gemini 2.5 Flash Image (Nano Banana) and FLUX.1 Kontext [pro], treating them as collaborators, not replacements. They can spark ideas in seconds, but the real creative authorship still happens in Photoshop, where you take back control through layers, masks, and adjustments.

My holiday card is the perfect example.

It started from one Thanksgiving family picture — and turned into multiple finished card concepts.

So here’s the challenge I’m inviting the community to try:

Run your next concept twice.

👉 Once with AI (your partner model of choice)

👉 Once fully by hand

Then compare your speed, control, and ownership. If you share your results, use the Structure Formula:

[Adobe Firefly] + [Partner Model] → #Tag both

Partner models include: Gemini 2.5 Flash (Nano Banana), FLUX Kontext, Google Imagen, Google Veo, GPT Image, Ideogram, Runway, Pika, Luma Ray, etc.

Excited to see what you discover.

Who’s really driving the brush?

[Gord@APL]

On balance, I think the implementation of AI by Photoshop is more productive than the use of AI generally.

There are already studies that seem to indicate that use of AI - at least where crafting text is concerned - is actively dumbing us down and making it more difficult to work independently of AI in future, and LLMs' propensity for hallucinating limits their worth.

Photoshop's usage of AI to expedite and simplify tasks that would previously have required hours of painstaking work with the clone brush has been a fantastic improvement to work-fluidity. This feels like the ideal, if not wholly Utopian way to employ AI: it's essentially taking on the drudgery of what a human would have to do, but quicker and more efficiently. I've found it invaluable for extending backgrounds and removing unwanted elements - particularly clearing a patch in an image to ensure the legibility of image captions. I'm happy to be dependent on AI for such tasks because it leaves me more time for the more important creative tasks.

Personally, I'm rather more sceptical of AI as a means of generating visual concepts... If they're used to draft a basis to be developed into a concept to present a client, for example, that's one thing... actually presenting the AI concept feels lazy and disingenuous.

The old adage of "a picture is worth a thousand words" takes on a whole new meaning, because a well-crafted photograph/visual, which may have required models/actors, set-dressers, lighting techs, etc. will invariably be more costly than its closest AI-generated analogue, but it will be more precise and 'real' than something created through a thousand words' worth of prompt.

[Amybeth M. ACI G7]

Insightful reflection — you highlight the balance between AI as a tool that enhances creativity versus one that risks replacing authentic human effort @Gord@APL

[Gord@APL]

That's going to be the real battle, I think...

Those outside the creative sphere are more likely to take on that risk, since it's an area they frequently don't fully understand (or respect, sadly) the work/skill/experience that goes into creativity. If AI offers a cost saving, that's all they'll need/want to know about it. If it also saves time, so much the better... But it does rather make one think of the 'Iron Triangle': if you get it fast and cheap, it ain't gonna be good.

The creatives, meanwhile, will hopefully embrace the timesaving aspects, but remain cautious of 'overusing' AI because they're more inclined to keep their skills finely tuned. I'm looking forward to the next AI innovations brought into Photoshop.

[Amybeth M. ACI G7]

Sometimes, depending on the end use, compositing projects are simply faster when I prompt an asset instead of digging through endless folders “searching” for the right one. It’s not every project — it really depends — but there are plenty of times where generating a clean, purpose-built piece is quicker and more efficient than sorting through a mountain of image assets

[Gord@APL]

I absolutely agree there - I had a situation recently where a client misunderstood what they'd booked into one of the publications I work on (a templated 'advertorial') and provided a low res visual for a branded advertisement. The only assets they provided were photographs that took up maybe a quarter of the page space.

Because they weren't able to provide high res versions of the background and decorative elements, and my employers doesn't subscribe to any stock image libraries, I ended up using Photoshop's AI features to upscale and repair the low res background, and generate the missing decorations. They were... passable... looked good at a glance, and saved a whole lot of time.

[derekwatson]

I agree, I love the time saving it gives me in Photoshop and Turntable in Illustrator will aslo save time and effort.

I am against using AI for ideation (I do dislike that word for some reason) as thats the creative part and who knows where those 'ideas' have come from.

[Amybeth M. ACI G7]

Totally with you. The efficiency gains are fantastic — Photoshop and Turntable are legit time-savers — but letting AI cough up “ideas” feels like borrowing someone else’s sketchbook with no clue who drew what. Keep the creativity human, use the tech for the grunt work.

[Amybeth M. ACI G7]

I am r curious as to hear– on"average" What is the balance in your workflow? Which way do the scales tip? Or like myself, is it project based? #adobefirefly #Nanobana #FluxKontetPro

[derekwatson]

Definitely project based, I am not using any gen AI from partner models in client work but am using them extensively for testing purposes. I do find it can take longer tweaking and editing generated content than adding a stock image and editing... The sourcing the right image is the time consuming bit.

[Rollan Banez]

Very insightful video Amybeth, we are still the navigators of creativity, some may view it as trick because of the lack of understanding of the actual workflow that involves the creation such as decision making, and other factors. Access to partner models can give us more ways to experiment and think beyond the box. Human intelligence + technology is the way for us to improve going forward. 

[Amybeth M. ACI G7]

Absolutely. We’re still the ones steering the ship—AI just hands us a few new sails. What people call a “trick” is usually just a gap in understanding the actual craft: the decisions, the intent, the workflow, the why behind every move. Partner models don’t replace that; they expand the sandbox. The real magic happens when human intelligence and technology tag-team the process. That’s how we level up, not lose ourselves

[barbara_a]

It's making us both more dependent and more efficient. I would not want to go back,

[Amybeth M. ACI G7]

Perfectly said!

[fbeaume]

Personally, I'm not a fan of AI invading every area of graphic design, but I have to admit that for certain stages of the creative process, some AI-based tools can help us be more efficient!
In my opinion, newcomers to the creative field are more likely to become dependent if “traditional” tools and knowledge are neglected in favor of AI tools. Remember that for now, AI is still dependent on humans... Let's make sure it stays that way and that AI helps humans build the world of tomorrow rather than letting it generate it for us... And who knows, in the near future, 100% AI-free could even become a trendy thing... What do you think?

[D Fosse]

AI is still dependent on humans...

By @fbeaume

Actually, there is increasing evidence that AI is dependent on AI. Few people realize this, but by the latest estimates, roughly 50 % of all content on the internet is now AI-generated...

I don't understand why more people aren't talking about this.

[Amybeth M. ACI G7]

I love the efficiency boost—Turntable, Generative Fill, all those “thank-you-for-saving-my-Saturday” features—but they’re tools, not a replacement for the creative muscle memory we spent years building.

The danger isn’t AI itself, it’s skipping the fundamentals. If you don’t know how to color-correct by numbers, build a clean path, or fix a file that goes sideways on press, you’re basically handing the steering wheel to a tech that still needs training wheels.

Honestly? I think you’re right—“100% AI-Free” could absolutely become a flex.

[Roni_Erric6360]

Thanks for opening up this discussion. Your approach of treating these tools as creative partners instead of replacements makes a lot of sense. Running a concept both digitally assisted and fully by hand is a great exercise for understanding how much control and authorship we keep in the process. I like the idea of comparing speed and decision-making side by side. Your holiday card example really shows how these workflows can complement each other rather than compete. Looking forward to trying this challenge and seeing how others interpret it.

[Amybeth M. ACI G7]

I cannot wait to see the results of your personal challenge! Please share !

[christianot40873293]

I find it fascinating how each model has its own characteristics, as if they were several collaborators, each with their own talent, and by combining them all, you have a multi-talented team.

The creative freedom that Photoshop and Adobe bring to us designers is fascinating, enabling all these technologies to be united within an efficient platform.

How can you not fall even more in love with Photoshop?

[Amybeth M. ACI G7]

I agree, one tool, one canvas that can complete a variety of creative tasks!

[Valdair Leonardo]

That is an excellent point! Indeed, there is a very fine line between the efficiency that AI offers and the dependency it can generate.

Your observation regarding prompt writing is the perfect proof of this. If we delegate more and more basic tasks, we end up losing foundational knowledge and the ability to give precise and effective commands. The quality of the AI's output depends directly on our ability to articulate what we truly want.

When AI allows us to automate the "how," we risk forgetting the "why" and the "what." If we use AI to think less about the structure of a command, we become less adept at guiding the tool when the task becomes complex or demands creative nuances that only a refined prompt can achieve. In essence, AI amplifies our intention, but if our intention becomes vague due to the underuse of our skills, the result will be mediocre. Thank you very much for this valuable insight!