A good documentary about Datacenters.
Actually, this Gemini Omni looks serious. The video the host created using Gemini Omni looks good and I can definitely see people who create video presentations subscribing to this via their google profile. As I mentioned in an earlier post Google definitely has an advantage over most other competitors as their current business model generates very high levels of profit.
This is definitely a very strong AI product that millions will have a sensible and effective use for.
I also really like Google’s marketing on this, this is definitely marketed as a tool which normal people and film/picture companies can use, it would seem google’s approach is to use AI to assist people rather than replace them.
I would think some production companies would lose some business due to this, on the other hand there are still many people who would want their video’s produced by people. Most people still have the viewpoint that human experience in a particular field is very much valued.
When people say this I always ask them what experience they think the training dataset contains. Why should I rely solely on one human’s set of experiences, when an AI can be trained on hundreds or thousands of professional testimonials and training materials.
What you are saying may be fact, and I suppose the person who has a great understanding of this could well choose the AI option.
For the people who don’t quite understand this they will still hire a producer (where the producer may use Gemini Omni as part of their toolset) if they want to make a video.
Also the person who has commisioned the video may not have time to fiddle around with Google Gemini Omni for 12 hours and would simply prefer to get on with what ever business they do and pay someone to handle it. I suspect many video producers will find Omni useful as well as people who have the time to try their hand at making a video.
Lets say if an AI robot that paints walls is created. I think I would still choose the human with 30 years experience to paint my house, then if this same painter has the AI painter robot to do some of the painting I would be cool with that.
Again, those who have a general understanding know that painting involves removing plugs, sanding down, applying barriers. there is also a technique to painting with sanding after the first coat. My point is that AI will still have difficulty due to people still having an old fashioned mentality that a person is better suited for a job rather than a robot, Yes, An AI built on 1000 people’s experience is great but in some aspects of work it still makes perfect sense to hire a human.
Answer me honestly, Who would you want fixing your electrical wiring and natural gas pipes in your home? A human with 25 years experience or an AI trained on 1000 engineers?
I definitely think some businesses do have to be careful with AI adoption, the assumption that everyone will be okay with an AI operator when in fact they want to speak/work with a human might actually make customers leave and subscribe to a company where they can speak/work with a human.
For now we don’t have any AI capable of doing these things, so obviously I will pick the human.
But once we have robotic workers, I wouldn’t object to using them. My objection is solely based on the fact that this technology does not yet exist.
As a better example, let’s take something AI can actually do. I could hire a human artist to do portraits of my character for EVE. But I use AI because it’s faster, cheaper, and produces superior results.
For your edit, I do prefer speaking to a human in customer support situations, but that’s simply because the human is actually able to solve problems, whereas the AI will generally only be allowed to parrot the first level script at you. Yes I have rebooted, yes I have reinstalled, please let me talk to a human so they can clear your database’s cached key on me and let me actually connect.
The current AIs only produce statistically relevant averages, so whatever you need that it’s not an average, you’re out of luck with an AI, specially if you don’t know that what you need would be not-average.
And there’s another issue, too: averages don’t exist in the real world.
You know, the average colour is either a shade of gray (for additive colours) or a shade of brown (for subtractive colours), but chances that someone needs exactly the average grey or the average brown are incredibly slim: the average colour does not exist and people use either specific shades or combinations of shades, none of which are “average”.
Same goes for the average (text), average (image) or average (sound), thus AIs essentially churn out something that is “OK-ish” but not the real deal of what the user needs. Of course, you can hire a human and get a bad product too, but also an out of ordinary human will give you an out of ordinary product (for good or for bad) and that might be exactly what you needed without knowing it.
AIs are tools, not minds. Anyone can saw pieces of wood and join them into a piece of furniture. But someone’s whose dealt with 100 different custom furniture sets for 100 different custom environments for 100 different customers will NOT produce an average: will produce exactly what the customer needs, or something close enough, a product of their expert mind who know not just “what” or “how” but also “why”.
That’s what my own experience points: I’m in a niche that’s been “killed by internet” for 20 years and yet here I am, still doing “the same you can do at home with a computer” and “the same an AI could do for you”, which turns not to be what my customers need.
Because I’m not an average and my skill set is replicable to a point, but not “averageable”.
2 Likes
Hmmm, we’re going to have to negotiate on the type of example used, could we stick to an example where the end result is promoting a product or earning money?
Okay, so let’s take generating music. I could make my own music, I’m a good pianist, a decent flutist, and an adequate singer, and I know the basics of writing sheet music and lyrics. But with AI I just tell it the lyrics and the genre of song I want, hit generate a bunch of times, and pick the best result. I can then upload that to Spotify through a distributor and et voilá, I’m earning money by selling a product.
Which is exactly what people will want to pay YOU for. they want to hear precisely how YOU play the flute which seems to be the general reason why people listen to lots of different artists. How on earth are people going to hear your unique way of playing that they haven’t heard anywhere else?
Also, you actually have to specify that a song is AI generated, otherwise you are misleading your spotify customers. Follow the process and see how many sales you get if it is clearly marked AI generated, then complare that to the sales of your Original and unique material.
Back to your original example, if CCP used AI aspects in character generation that would be fine. simpily because Eve online was handcrafted by humans, it would be used as a tool by CCP to enhance part of their game it would not be taking over the whole enire game.
As I said in my previous answer, what you’re clashing on is the non-averageable nature of your tech issues. And surely you’re lowering your expectations to AI-can-do in terms of your custom portraits… because the EVE online character creator dataset is too small to be successfully replicated as an average.
If I wanted to make money, I’d focus on rap, since that seems to be the biggest genre at the moment (to my personal bafflement). I can write rap bars, but I can’t rap personally. AI can. If we were going to compare hypothetical sales figures between “Anna Rap” and “AI Anna Rap”, the AI will do much better simply because its tracks will be listenable.
No you’re not, because your tune just got drowned among the other million tunes similarly generated today, of which a few hundreds will sound exactly like yours and probably some will even have similar lyrics because you had to confine your creativity to the averageable envelope of the AI. You would do “what was good enough” but not “the exceptional”.
And yes, you might create a turd. Most human creations are of questionable quality, but here’s the thing… we create for ourselves. My AI slop images are exactly that, slop. But I can watch them and turn them into the real deal in my mind, and the act of creating the prompt is satisfactory in itself…
Hmmm, this might be an insult to rap music. If you check the origins of rap music it started as a way for people to express an opinion about their struggles, many still follow this and I’m not sure that an AI rap artist would sell well.
One of the first commecial rap songs:
This song has been legally sampled by so many artists, it seemed to set the standard.
1 Like
(By courtesy of Ishtanchuk): Rap’s gateway into mainstream was this piece by Blondie, which even mentioned Flash himself (and he acknowledged the difference that made in his career and how Blondie’s intervention moved rap out of club culture in the Bronx):
Blondie - Rapture (1981)
2 Likes
Indeed, there is maximum respect for the Blondie - Rapture;
1 Like
Are we off topic yet?
Still on topic, someone claimed that they could make an AI rap artist and it would sell well, I’m not so sure.
Back on topic… guess who’s retirement fund is gonna buy head over heels the scammiest IPO ever scammed because your money is handled by bots…
Like, seriously…
What you need to know about buying in the SpaceX IPO as “retail”: don’t.
I didn’t claim it would sell well, I claimed it would sell better than if I tried to rap myself.
LOL, i retract my previous statements as I just found this;
Xania Monet: An entirely AI-generated R&B “artist” whose songs have climbed the charts, securing a $3 million record deal with Hallwood Media.
Velvet Sundown: A fully AI-created band that has achieved millions of streams on major platforms
1 Like