ChatGPT accords

[Perfumer_]

I decided to ask ChatGPT for some perfume accords, it actually comes up with some numbers haha. I have no idea where it gets the info. Pretty cool:

Sandalwood

• Cedarwood oil: 40%
• Amyris oil: 30%
• Vanillin: 20%
• Vetiver oil: 10%

Rose

• Geraniol: 50%
• Citronellol: 25%
• Phenylethyl alcohol: 15%
• Beta-damascenone: 5%
• Rose oxide: 5%

Plum

• Benzaldehyde: 50%
• Ethyl maltol: 30%
• Gamma-decalactone: 10%
• Allyl heptanoate: 5%
• Methyl anthranilate: 5%

Tea

• Iso E Super: 40%
• Galaxolide: 20%
• Hedione: 20%
• Bergamot oil: 10%
• Blackcurrant bud absolute: 5%
• Cetalox: 5%

Leather

• Iso E Super: 40%
• Birch tar oil: 25%
• Castoreum absolute: 15%
• Labdanum absolute: 10%
• Cedarwood oil: 5%
• Vanilla absolute: 5%

Jasmine

• Jasmine absolute: 50%
• Ylang ylang oil: 20%
• Bergamot oil: 10%
• Neroli oil: 10%
• Sandalwood oil: 5%
• Vanilla absolute: 5%

Pear

• Iso E Super: 40%
• Ethyl maltol: 25%
• Allyl amyl glycolate: 15%
• Dimethyl benzyl carbinyl acetate: 10%
• Methyl octine carbonate: 5%
• Alpha ionone: 5%

Oud

• Agarwood oil: 40%
• Patchouli oil: 20%
• Cedarwood oil: 15%
• Sandalwood oil: 10%
• Labdanum absolute: 10%
• Cypriol oil: 5%

Marine

• Calone: 40%
• Ambroxan: 20%
• Sea salt: 15%
• Grapefruit oil: 10%
• Cedarwood oil: 10%
• Jasmine absolute: 5%

Sea salt

• Methyl salicylate: 40%
• Ethyl maltol: 20%
• Iso E Super: 15%
• Sea salt: 10%
• Lemon oil: 10%
• Lavender oil: 5%

You can try it out, I just ask it:
"how to make a "x" accord for perfume."
Results are interesting....

 

[David Ruskin]

Have you made any of these? Some of the numbers look a little strange. You would probably have difficulty in getting some of the solid materials to dissolve, at the levels present in some of these accords. By the way, what is "sea salt"?

 

[RomanB]

Have you made any of these? Some of the numbers look a little strange. You would probably have difficulty in getting some of the solid materials to dissolve, at the levels present in some of these accords. By the way, what is "sea salt"?

ChatGPT is a program that is able to plausibly answer questions, but it doesn’t care about their correctness. If you will say to it that 20% vanillin is too much, it will correct itself to please you, but the new answer won’t be more real.

 

[Alex F.]

Of course the numbers look strange. This was generated by "artificial intelligence" (which isn't intelligent enough yet to recognise that sea salt is not a perfumery ingredient, or that birch tar is an extremely powerful material, ...).
The number of discussions about it on this forum shows that there's a lot of fascination for this new technology. If it spikes your creativity, fine. Show some results of your experiments. Otherwise, posting these outpourings is useless.

 

[Big L]

From the last time that we had this discussion, here is my solution to the numbers problem.

 

[Casper_grassy]

Birch Tar @25% hahahaha

 

[David Ruskin]

Would it not be as effective to use dice? Or pick from a list of ingredients at random? Or just use the skills of a perfumer.

There have been several posts devoted to this subject and I still do not see the point.

 

[David Ruskin]

ChatGPT is a program that is able to plausibly answer questions, but it doesn’t care about their correctness. If you will say to it that 20% vanillin is too much, it will correct itself to please you, but the new answer won’t be more real.

Well I can do that, and so can anyone. If you ask me a question I think I can make good stab at a plausible answer, especially if I don't care if that answer is correct. What is the point?

 

[RomanB]

Well I can do that, and so can anyone. If you ask me a question I think I can make good stab at a plausible answer, especially if I don't care if that answer is correct. What is the point?

The point is that the program is already good enough to to take millions of jobs right when it will be commercially available. You will call a bank and it will talk to you, it will answer all your questions, it will propose solutions and will even joke if you want (it is capable of creating original jokes). This isn’t in the future, it happens right now, you can register and test it yourself. People are shocked by the program’s capabilities and try do find what it can do already, without any more tweaks. I know that it already wrote a Univerity’s thesis and it was successfully defended. It passes job interviews, gets nurse licenses, et cetera.

 

[Casper_grassy]

“How to win the Lotto?”

 

[David Ruskin]

This other programme of which you speak, seems far more capable than the ones we have been shown here. Do these other, more efficient, programmes have a better sense of ethics or do they also not care about the truth?

I have had "conversations" on line with my bank, and for the majority of occasions found them to be highly unsatisfactory. There is a huge difference between trying to explain to a computer programme, and explain to a human being. No doubt things will improve; at the moment I am less than convinced of their ability.

And the bank programme has nothing to do with the creation of fragrances, or music, or art. So far, I am not impressed by the ability of the programmes shown here to create anything of worth.

 

[RomanB]

This other programme of which you speak, seems far more capable than the ones we have been shown here. Do these other, more efficient, programmes have a better sense of ethics or do they also not care about the truth?

I have had "conversations" on line with my bank, and for the majority of occasions found them to be highly unsatisfactory. There is a huge difference between trying to explain to a computer programme, and explain to a human being. No doubt things will improve; at the moment I am less than convinced of their ability.

And the bank programme has nothing to do with the creation of fragrances, or music, or art. So far, I am not impressed by the ability of the programmes shown here to create anything of worth.

As of visual arts, search for “Midjourney”. It creates images from your descriptions. You can ask it to paint “a six-legged perfume bottle with four butterfly wings, sitting under a sakura tree, with a Godzilla in the background, in style of Salvador Dali”, and it will paint it to you in many variants. Change the style and it will paint the scene as impressionist or pre-raphaelite. Tell it to add a moon made of cheese, and so will be done immediately.

It is bad at painting fingers and football players, but very impressive overall.

 

[David Ruskin]

As of visual arts, search for “Midjourney”. It creates images from your descriptions. You can ask it to paint “a six-legged perfume bottle with four butterfly wings, sitting under a sakura tree, with a Godzilla in the background, in style of Salvador Dali”, and it will paint it to you in many variants. Change the style and it will paint the scene as impressionist or pre-raphaelite. Tell it to add a moon made of cheese, and so will be done immediately.

It is bad at painting fingers and football players, but very impressive overall.

I've used it; it's crap.

 

[David Ruskin]

I'm being unfair. I'm sure Midjourney is technically very clever, it's just that I don't think the results are very good. Firstly, it doesn't paint anything; it makes an image that is shown on a screen. Secondly, I don't want to work in the style of Dali, I want to find my own style and work in that. Both this programme and the ones for perfumery shown to us here, seem not to factor in creativity, originality or the personal touch. Mixing together a bunch of stuff, for whatever reason, cannot better the experience and creativity of a human. Not yet anyway.

 

[the_good_life]

This other programme of which you speak, seems far more capable than the ones we have been shown here. Do these other, more efficient, programmes have a better sense of ethics or do they also not care about the truth?

I have had "conversations" on line with my bank, and for the majority of occasions found them to be highly unsatisfactory. There is a huge difference between trying to explain to a computer programme, and explain to a human being. No doubt things will improve; at the moment I am less than convinced of their ability.

And the bank programme has nothing to do with the creation of fragrances, or music, or art. So far, I am not impressed by the ability of the programmes shown here to create anything of worth.

I've experimented with ChatGTP alot, because its capabilities are of concern to educators. While it's neither creative nor intelligent it can already do an excellent job at performing routine tasks such as creating standardized texts from data input, which, as Roman has pointed out, will evaporate millions of office/screen jobs which consist of doing precisely that. Writing webcontent, writing rote top 40 songs, creating standard perfume formulas, initial diagnoses, rote architecture will all be hands down once these things are fully linked up to gloabal databases. After three tries ChatGTP outperformed every one of my students on a typical high school essay writing task. In fact, you can tell it to create an essay as written by a mediocre or bad student and it will simulate those pretty well, too. Meaning that unsupervised essay writing in school is over.

 

[Grant]

Off topic, but it puts me out of a job too:

Create a news story about the following new release for a perfume news website:
Name: Dorian’s Fougère
Top Notes: bergamot spike lavender petitgrain bigarade mastic leaf
Heart Notes: carnation geranium hawthorne hay absolute sambac jasmine neroli tonka bean absolute
Base Notes:balsam mysore sandalwood patchouli labdanum saffron absolute oakmoss vintage musk vintage civet Perfumer: Darran Alan
Sizes and Price: 50ml Parfum ($155.00 USD) 10ml Parfum($37.00 USD)
Inspired by: Oscar Wilde's novel "The Picture of Dorian Gray" (maybe add a synopsis of the novel)

---

Perfume lovers, get ready to indulge in the new release from Darran Alan, "Dorian's Fougère," inspired by Oscar Wilde's famous novel "The Picture of Dorian Gray."

This exquisite fragrance boasts a blend of bergamot, spike lavender, petitgrain bigarade, mastic leaf, carnation, geranium, hawthorne, hay absolute, sambac jasmine, neroli, tonka bean absolute, balsam, mysore sandalwood, patchouli, labdanum, saffron absolute, oakmoss, vintage musk, and vintage civet.

This luxurious scent is available in two sizes, 50ml and 10ml parfum, priced at $155.00 USD and $37.00 USD, respectively. The novel "The Picture of Dorian Gray" explores the dangers of vanity and the desire for eternal youth, which is reflected in the scent's unique blend of vintage and modern ingredients.

Experience the essence of this classic tale with "Dorian's Fougère," the latest addition to Darran Alan's fragrance collection.

 

[mnitabach]

I've experimented with ChatGTP alot, because its capabilities are of concern to educators. While it's neither creative nor intelligent it can already do an excellent job at performing routine tasks such as creating standardized texts from data input, which, as Roman has pointed out, will evaporate millions of office/screen jobs which consist of doing precisely that. Writing webcontent, writing rote top 40 songs, creating standard perfume formulas, initial diagnoses, rote architecture will all be hands down once these things are fully linked up to gloabal databases. After three tries ChatGTP outperformed every one of my students on a typical high school essay writing task. In fact, you can tell it to create an essay as written by a mediocre or bad student and it will simulate those pretty well, too. Meaning that unsupervised essay writing in school is over.

LLMs are very good at creating mediocre median drivel, which makes perfect sense given how they work. What they can't do is discern the causal structure of actual reality. This is why LLMs fail abysmally at anything that hinges on small but critical factors that have an exponentially large effect on outcomes. Of course, as was pointed out above, a huge amount of commercially & educationally relevant text is precisely "mediocre median drivel", which is why LLMs will have a big impact on society. But they can't even come close to fooling experts in domains where expertise matters.

 

[David Ruskin]

I've experimented with ChatGTP alot, because its capabilities are of concern to educators. While it's neither creative nor intelligent it can already do an excellent job at performing routine tasks such as creating standardized texts from data input, which, as Roman has pointed out, will evaporate millions of office/screen jobs which consist of doing precisely that. Writing webcontent, writing rote top 40 songs, creating standard perfume formulas, initial diagnoses, rote architecture will all be hands down once these things are fully linked up to gloabal databases. After three tries ChatGTP outperformed every one of my students on a typical high school essay writing task. In fact, you can tell it to create an essay as written by a mediocre or bad student and it will simulate those pretty well, too. Meaning that unsupervised essay writing in school is over.

Why does this fill me with great pessimism for the future? Not to mention a slight concern at, either your students ability to write good essays, or your ability to teach them. Who defines "mediocre" or "bad student"? How do you teach a computer to write badly, or in a way that you would mark poorly?

 

[David Ruskin]

Off topic, but it puts me out of a job too:

Create a news story about the following new release for a perfume news website:
Name: Dorian’s Fougère
Top Notes: bergamot spike lavender petitgrain bigarade mastic leaf
Heart Notes: carnation geranium hawthorne hay absolute sambac jasmine neroli tonka bean absolute
Base Notes:balsam mysore sandalwood patchouli labdanum saffron absolute oakmoss vintage musk vintage civet Perfumer: Darran Alan
Sizes and Price: 50ml Parfum ($155.00 USD) 10ml Parfum($37.00 USD)
Inspired by: Oscar Wilde's novel "The Picture of Dorian Gray" (maybe add a synopsis of the novel)

---

Perfume lovers, get ready to indulge in the new release from Darran Alan, "Dorian's Fougère," inspired by Oscar Wilde's famous novel "The Picture of Dorian Gray."

This exquisite fragrance boasts a blend of bergamot, spike lavender, petitgrain bigarade, mastic leaf, carnation, geranium, hawthorne, hay absolute, sambac jasmine, neroli, tonka bean absolute, balsam, mysore sandalwood, patchouli, labdanum, saffron absolute, oakmoss, vintage musk, and vintage civet.

This luxurious scent is available in two sizes, 50ml and 10ml parfum, priced at $155.00 USD and $37.00 USD, respectively. The novel "The Picture of Dorian Gray" explores the dangers of vanity and the desire for eternal youth, which is reflected in the scent's unique blend of vintage and modern ingredients.

Experience the essence of this classic tale with "Dorian's Fougère," the latest addition to Darran Alan's fragrance collection.

This seems to be as empty of content as most modern marketing campaigns, so I guess it is doing a good job! Nothing original though.

 

[Perfumer_]

Of course the numbers look strange. This was generated by "artificial intelligence" (which isn't intelligent enough yet to recognise that sea salt is not a perfumery ingredient, or that birch tar is an extremely powerful material, ...).
The number of discussions about it on this forum shows that there's a lot of fascination for this new technology. If it spikes your creativity, fine. Show some results of

Yes, the point wasn't to create the formulas, but rather interesting to think about where it derived the info from. If you google Plum accord there is no real results, nothing on thegoodscents demo formulas... But it managed to come up with some sort of results. It must be using this forum in fact for some of its knowledge querying the DIY section for info, among other sites. It's interesting technology, the "outpourings are useless" sure, but give it some time until AI can start making better formulas than a human. (Matter of time?) It's up for debate.