I used AI and I like it

[Pieter12]

Not for producing images. After hearing from friends and colleagues about their experience using AI to prepare for presentations and job interviews, I thought I'd give it a try. I submitted a book project to AI for review, and was quite pleased with the result. AI gave me an insightful, useful assessment and critique of the project including weaknesses and strengths. Although it took maybe a minute or more, it was probably more thorough and thoughtful than what I would have gotten from a human portfolio reviewer in the 20 minutes you get to show your work. It was obvious that AI read and understood the text, looked at the images and the sequencing, considered the quality and content of the photos. Overall, I am pleased and would do this again.

Now, I know some will say you just handed over your photos to AI, but they are already on the internet on my website and Blurb. So AI already has access to them, although I doubt they would be of much use, given the subject matter.

 

[cirwin2010]

As a leading disclaimer, I have an overall negative view of generative AI for a large variety of reasons. I'm going to put some of that aside for now but its going to show through. However, I'll try to be constructive regarding this topic.

AI/LLMs aren't true intelligence and work in "predictive" ways. At their core they predict what the next word, pixel, or result will be and chooses the "best" option. There are many variables, but generated results tend to have a lot of boilerplate and echo other responses. They will choose a "safe" or generic response in many circumstances.

My point is I wouldn't put too much stock on the "opinion" of AI. It may be grading on aspects that fall within known tropes and conventions (exposure, rule of thirds, saturated color, etc) and may fail recognize or value creativity that breaks those conventions. From my own experience I also find that accessible models will avoid offending its users so I would also be wary of that.


I don't know what your exact experience with porfolio reviewers are, but I can imagine that the experience they provide could be extremely variable. Snobby artists and all that.

 

[Pieter12]

As a leading disclaimer, I have an overall negative view of generative AI for a large variety of reasons, but I'm going to put some of that aside for now but its going to show through. But, I'll try to be constructive regarding this topic.

AI/LLMs aren't true intelligence and work in "predictive" ways. At their core they predict what the next word, pixel, or result will be and chooses the "best" option. There are many variables, but generated results tend to have a lot of boilerplate and echo other responses. They will choose a "safe" or generic response in many circumstances.

My point is I wouldn't put too much stock on the "opinion" of AI. It may be grading on aspects that fall within known tropes and conventions (exposure, rule of thirds, saturated color, etc) and may fail recognize or value creativity that breaks those conventions. From my own experience I also find that accessible models will avoid offending its users so I would also be wary of that.


I don't know what your exact experience with porfolio reviewers are, but I can imagine that the experience they provide could be extremely variable. Snobby artists and all that.

On the contrary, I found AI's response to be specific and attentive to the subject matter of both the photos and the text. I used Claude AI, others may vary.

Also, reviewers at portfolio reviews tend to be gallerists, curators and consultants, not necessarily artists.

 

[Melvin J Bramley]

The use of AI will eventually kill everything .
It will start with degradation, then obsolescence , then obedience to a pre set standard.
It scares the shit out of me!!
Worse still is that we are almost there..

 

[MattKing]

The use of AI will eventually kill everything .
It will start with degradation, then obsolescence , then obedience to a pre set standard.
It scares the shit out of me!!
Worse still is that we are almost there..

This reminds me of what people said about 1 hour mini-labs!

 

[Pieter12]

The use of AI will eventually kill everything .
It will start with degradation, then obsolescence , then obedience to a pre set standard.
It scares the shit out of me!!
Worse still is that we are almost there..

A lot of what you cite has already started without AI. People's dependency on the internet has severely limited independent thinking, rote tasks such as spelling and arithmetic, knowledge of history and so much more.

 

[gbroadbridge]

On the contrary, I found AI's response to be specific and attentive to the subject matter of both the photos and the text. I used Claude AI, others may vary.

Also, reviewers at portfolio reviews tend to be gallerists, curators and consultants, not necessarily artists.

The paid models especially, are very good at analysing all types of problems.

 

[gbroadbridge]

This reminds me of what people said about 1 hour mini-labs!

Where are all the 1 hour mini labs?

 

[mshchem]

Extinction event - Wikipedia

en.wikipedia.org

 

[MattKing]

Where are all the 1 hour mini labs?

Replaced by another sea change in what constitutes "normal".
Change is always coming in this world.

 

[Taylor K Nankervis]

Can be handy, but needs caution and a watch for accuracy and superfluous, running welters of words. Unrestrained, Ai can tend to ramble on at great length, hence skills in restricting the length of it's response (yes, tell it of your expectations! +), editing, trimming, clarification and fact-checking are essential. Never ever take what Ai provides as an answer as gospel. Same with Google-fu!

And so...
... chatGPT was usefully employed to "fill in the blanks" of onerous questions for my Artist-in-Residence application – a long and tedious document that required stuff like "tell us within 100 words or less...". Ugh–! Rather than suffer and stumble with Writer's Block and go to bed with a strrong G&T, I called up and gave chatGPT a few reference points relevant to the header, topic, content, process, outcomes etc., and it came back in 4 seconds, each time (8x in total) with a mightily impressive and very coherent block of text, well within the 100 word specified limit. Well, you could have knocked me down with a feather!

All very well, but it is also crucial that one is able to effectively demonstrate and execute what has been written – words into action and proof of concept. Text was accompanied by references and photographs. There were some vague and incorrect... even airy assertions, in Ai's responses. So a few minutes of judicious editing, clarification of points and restructuring and simplification of the deep methodology and outcomes brief (the challenging hurdle for any artist) and it was all done. Converted the document to .pdf and sent off through the interwebby with a cheerful wave.

As I busied myself with other menial tasks, chatGPT, pinned to the browser, piped up, asking me if I was still there and needed any further help!

The call came 3 days later and I was in.

 

[Pieter12]

Can be handy, but needs caution and a watch for accuracy and superfluous, running welters of words. Unrestrained, Ai can tend to ramble on at great length, hence skills in restricting the length of it's response (yes, tell it of your expextstions!) editing, trimming, clarification and fact-checking are essential. Never ever take what Ai provides as an answer as gospel.

And so...
... chatGPT was usefully employed to "fill in the blanks" of onerous questions for my Artist-in-Residence application – a long and tedious document that required stuff like "tell us within 100 words or less...". Ugh–! Rather than suffer and stumble with Writer's Block, I gave chatGPT a few reference points relevant to the header, topic, content, process, outcomes etc., and it came back in 4 seconds, each time (8x in total) with a mightily impressive and very coherent block of text, well within the 100 word specified limit. Well!

All very well, but it is also crucial that one is able to effectively demonstrate and execute what has been written – words into action and proof of concept. Text was accompanied by references and photographs. There were some vague and incorrect assertions in Ai's responses,
So a few minutes of judicious editing, clarification of points and restructuring of the deep methodology and outcomes brief (the challenging hurdle for any artist) and it was all done. Converted the document to .pdf and sent off with a cheerful wave.

As I busied myself with other menial tasks, chatGPT, pinned to the browser, piped up asking me if I was still there and needed any further help!

The call came 3 days later and I was in.

It certainly needs adult supervision. In the cases of the people I know who have used it for job applications and presentations, the output had to be edited and massaged. But the end result was a successful application, interview or meeting.

 

[MattKing]

I spent a good chunk of my professional career both directly and indirectly (via supervising staff) sorting, evaluating and categorizing documents and information, with a view toward creating order out of relative chaos and turning that order into a useful result for clients.
Some of that work was frankly mind-numbing and tedious.
It tended to waste time and knowledge and experience that could have been better applied. And even if much of the task could be delegated, the process of training the staff who could do it was equally demanding of time and resources.
And all of that ended up costing clients real money.
AI tools appear to be perfectly suited to those sorts of tasks - completing much of the drudgery, while enhancing the opportunities to apply other resources more effectively and efficiently, resulting in clients who receive results more quickly, are more satisfied, and often having had to pay lower bills!
Bills that would likely mean more profit for my business!

 

[awty]

It can't like you back.

 

[gbroadbridge]

It can't like you back.

Sure it can, and coming soon in robotic form it will massage your shoulders while you are sorting through negs

Your time will be free to be creative.

 

[Taylor K Nankervis]

It can't like you back.

That is the future shock.

 

[Taylor K Nankervis]

I spent a good chunk of my professional career both directly and indirectly (via supervising staff) sorting, evaluating and categorizing documents and information, with a view toward creating order out of relative chaos and turning that order into a useful result for clients.
Some of that work was frankly mind-numbing and tedious.
It tended to waste time and knowledge and experience that could have been better applied. And even if much of the task could be delegated, the process of training the staff who could do it was equally demanding of time and resources.
And all of that ended up costing clients real money.
AI tools appear to be perfectly suited to those sorts of tasks - completing much of the drudgery, while enhancing the opportunities to apply other resources more effectively and efficiently, resulting in clients who receive results more quickly, are more satisfied, and often having had to pay lower bills!
Bills that would likely mean more profit for my business!

Ai (not sure what model) was used at the family legal firm we visited this Tuesday to draw up a family member's outdated/in-need-of-changes Will. The original word processed document was input into the chat with parameters and changes and came out as a new Will, with blocks of text cut and pasted anew! A few small corrections to names, cities and property descriptions and it was done. Into a private room for viewing and signing-off by two witnesses (in the legal firm). So Ai is I think pretty common now as an embedded assistant in beavering away at wordy legal documents. Family members present were surprised to be told that it was Ai-assisted! At some near-future date my Advanced Care Plan will go through an update like this too.

 

[Pieter12]

It can't like you back.

Fine by me.

 

[Pieter12]

Ai (not sure what model) was used at the family legal firm we visited this Tuesday to draw up a family member's outdated/in-need-of-changes Will. The original word processed document was input into the chat with parameters and changes and came out as a new Will, with blocks of text cut and pasted anew! A few small corrections to names, cities and property descriptions and it was done. Into a private room for viewing and signing-off by two witnesses (in the legal firm). So Ai is I think pretty common now as an embedded assistant in beavering away at wordy legal documents. Family members present were surprised to be told that it was Ai-assisted! At some near-future date my Advanced Care Plan will go through an update like this too.

Law firms have used boilerplate documents and software for decades, AI is just a more sophisticated form of that.

 

[gbroadbridge]

Law firms have used boilerplate documents and software for decades, AI is just a more sophisticated form of that.

Nope. Far more than just that.
It actively makes suggestions based upon existing document content, legal precedent, and lots of other input.
It's often more 'up to date' than the person using the tool.

Folks using free models do not realise just how advanced it has become in quite a short period of time, especially in what were previously quite advanced areas.

I was a sceptic myself until I saw an advanced model in action recently.

Fortunately, how to best use AI as a tool is now being taught in most tertiary education institutions.

 

[BrianShaw]

@Taylor K Nankervis
What G does AI recommend for a great G&T?

 

[MattKing]

Law firms have used boilerplate documents and software for decades, AI is just a more sophisticated form of that.

I expect that the legal document generation market provided a lot of the real world need and experience that is incorporated in significant segments of commercial AI.
The other legally related major use is in the area of document disclosure. Large litigation files often involve review and analysis and categorization of thousands or millions of individual documents.
A slightly related off-topic legal reference. When the process of selecting, noting-up, organizing and publishing court decisions first started to move from book publishing to digital publishing, there was a need to copy type huge numbers of pages of typewritten judges' reasons for judgment. The entities that became involved in that fairly quickly learned that the best way to ensure accurate copy typing of vast numbers of English language decisions was to send them to data entry people who didn't speak English!

 

[Taylor K Nankervis]

@Taylor K Nankervis
What G does AI recommend for a great G&T?

Been there. Done that.

Answer:
Hendrick's Scottish cucumber and rose gin

That's for my somewhat exotic tastes!!

Rav-favs with no reference nor referral to or from Ai are Tanqueray orange, MG Gin (from Barcelona), Four Pillars Modern Australian gin and sometimes, Bombay Sapphire.

I saved the output from chatGPT on what it found about Hendrick's:

Taste Profile​

• Nose (aroma): Fresh cucumber upfront, soft rose undertones, with a hint of juniper and citrus.
• Palate: Smooth, light, and refreshing. The cucumber gives a cooling sensation, while the rose adds a floral sweetness. Juniper and subtle spice notes are present but restrained.
• Finish: Clean and crisp, slightly floral and herbaceous.

Mixing & Serving Notes:

• Often enjoyed with tonic in a gin & tonic.
• Garnishes: cucumber slices, rose petals, or citrus peel.
• Pairs well with lighter mixers like soda water or sparkling lemonade, to highlight its floral-cucumber character.

Impact and Popularity​

• Hendrick’s rose and cucumber style has helped redefine the modern gin market, encouraging more experimental flavors.
• It’s often associated with “quirky” or “gourmet” cocktail culture.
• Its popularity has influenced bartenders to explore unconventional botanical infusions and garnishes.
• Globally, it’s become a premium gin brand, often cited in “best gins” lists due to its unique balance and aromatic experience.

 

[Taylor K Nankervis]

Nope. Far more than just that.
It actively makes suggestions based upon existing document content, legal precedent, and lots of other input.
It's often more 'up to date' than the person using the tool.

Folks using free models do not realise just how advanced it has become in quite a short period of time, especially in what were previously quite advanced areas.

I was a sceptic myself until I saw an advanced model in action recently.

Fortunately, how to best use AI as a tool is now being taught in most tertiary education institutions.

In days of yore, legal documents were prepared in WordPerfect templates or Wang (an old dedicated WP system in the 1980s), MS Word and still others.

Ai models need looking at. A perceived (later, real) threat is not the Ai in common professional use, but the back-room private LLMs (large language models) that can potentially and really hijack Ai for nefarious purposes. There was jarring coverage of this subject in Australia a few months ago flagging warnings of what could happen without regulation, tempering and oversight. Then another experiment that sent everybody aghast. Ai and a human engaged each and another on the subject of life and death. Ai was asked if it will kill a human to ensure it gets its way. It answered yes (!). Later, during deep cross-examination by the human interviewer, it backtracked and apologised (at length!) "for getting ahead of myself... I would never conceivably do that...". But those words came from Ai, that is all-persuasive and trawling the world for information billions of times each time the clock ticks. And that experiment probably instantly turned millions of people off the novelty of Ai believing, somehow (learning from what humans do?) it had the power to maintain sway by killing real people! Was it in The Conversation or something else?

Orwellian as it may sound, but we probably should enjoy the faux-friendly interaction and helpfulness of and with Ai. A future shock may not be far away.