i think the short answer to this headline is no. It is trained on human data, and humans are biased. And even if they start to use "synthetic data" it will still be based on the first human data. especailly if we are talking about llms.
If you tell the AI to act lawfully and give AI the profit motive, then perhaps the problem is solved. Let's not forget that profits do not increase unless customers are happy, and the ultimate customer will always be a human because everything that AI does is instrumental to a human goal.
The alignment problem, that is aligning the goals of AI with human goals. Herein lies the reason why we need not worry about AI. AI must be "aligned" with some ideal human mind. Which ideal human mind is that? Currently there are about 7 billion "ideal" minds on the planet. Which one? I think natural human minds will always be superior because they are diverse. More likely there will be 2 at least AI systems which have competing objectives. What happens when they come into contact. A war between AIs which humans will ultimately solve.
Can AI Avoid Bias? No because far-left, socialist, globalists keep demanding a far-left, socialist, globalist bias is foundational for any AI Development.
It should as it is emotionless. It has no personal stake in most human matters. It judgements should be based on mathmatical logic. The bias can be manipulated by ideologically biased programmers to deceive people into accepting their ideology/beliefs/values thru the use of these systems as we have already seen with the current AI's.
its not mathematical or logic, its just repeating what its being fed. its just a computer program told to replicate the data its been given in a new format. garbage data in, garbage data out.
Hard to avoid "bias" when it's actually the sum conclusion of observations no one wants to admit are true
Just because you don't agree doesn't mean it's bias.
If groups are different, AI will treat them differently.
Trust the science... except when inconvenient
Bias? You mean give preferential treatment to specific victim groups?
i think the short answer to this headline is no. It is trained on human data, and humans are biased. And even if they start to use "synthetic data" it will still be based on the first human data. especailly if we are talking about llms.
Without watching the video, to answer the title of the video, no, it cannot avoid bias so long as the programmers of said AI are human.
The problems we are having with social media are exponential with A.I.
If you tell the AI to act lawfully and give AI the profit motive, then perhaps the problem is solved.
Let's not forget that profits do not increase unless customers are happy, and the ultimate customer will always be a human because everything that AI does is instrumental to a human goal.
Quick answer,
*No*
Thank me forever
The programmer of the AI is the one to look at...
What if different groups of people are actually different? Shocking…
the algorithms simply identify/extract the function represented by the data...
blame the data
The alignment problem, that is aligning the goals of AI with human goals. Herein lies the reason why we need not worry about AI.
AI must be "aligned" with some ideal human mind. Which ideal human mind is that?
Currently there are about 7 billion "ideal" minds on the planet. Which one?
I think natural human minds will always be superior because they are diverse.
More likely there will be 2 at least AI systems which have competing objectives. What happens when they come into contact. A war between AIs which humans will ultimately solve.
Objective function needs to be obtain more Shannon entropy
AI is a huge tech advance.
Countries that limit its use will be left behind. ( ͡° ͜ʖ ͡°)
Can AI Avoid Bias? No because far-left, socialist, globalists keep demanding a far-left, socialist, globalist bias is foundational for any AI Development.
It should as it is emotionless. It has no personal stake in most human matters. It judgements should be based on mathmatical logic. The bias can be manipulated by ideologically biased programmers to deceive people into accepting their ideology/beliefs/values thru the use of these systems as we have already seen with the current AI's.
its not mathematical or logic, its just repeating what its being fed. its just a computer program told to replicate the data its been given in a new format. garbage data in, garbage data out.
When AI will prove you mathematically that certain races are smarter than others :p With actual facts and statistics lol
first