Covert Racism Found in Large Language Models
LLMs Are Not Intelligent They Are Stupid
Sitting at my desk today minding my own business relentlessly iterating my software business model to build a billion-dollar enterprise. When an article from Substack dropped on my desk. Titled Covert Racism in LLMs; incredibly important new paper with not-so-shocking (to me) revelations about technology.
The implications of the paper are serious and just more red flags about the danger of uncontrollable AI. Generative AI, both text and images can paint a picture of a world that amplifies bias in gender, race and beyond. The data that produce these toxic images and text are fundamentally fueled by Big Tech billionaires and the New Tech Aristocracy and their courtiers, aka VCs. This small group of mainly white males rules the universe and doesn’t care about the harm it produces, only about profit and control.
AI-generated stereotyping, inaccuracies, making stuff up, covert racism and lies a major emerging realities and challenges for the AI revolution and humanity as a whole.
The following is a sample from an excerpt found on X from the author’s thread — read it for yourself it’s a good explanation too of how LLMs work.
In short, everything having to do with “African American English” was negative and invoked many stereotypes. For example, on the legal side, ChatGPT is more likely to suggest that “defendants be sentenced to death when they speak African American English.” This is important because the machine is learning these negative assessments which further builds racism into the models and so too, society. Further exacerbating prejudice as AI decisions about people’s character, employment, and criminality reinforce ignorant stereotypes. Putting Black people at a tremendous competitive disadvantage. And as the widespread adoption of LLMs continues, the implications become more severe. And without any meaningful representation and ownership/equity in tech, there is no way for this to be stopped or even mitigated.
These language examples are only the tip of the iceberg because beneath the surface everything is driven by programmed language, by real people with inherent biases. So as I’ve been saying for a while now, LLMs are not intelligent they are stupid, and left unchecked, unregulated and in the hands of the Big Tech aristocracy, humanity will suffer.
LLM can’t empirically observe the world and human general intelligence is broad and includes all the five basic human senses: touch, sight, hearing, smell and taste. AI, on the other hand, is a narrow intelligence and dependent on being programmed with data. It can’t think on its own so not so intelligent after all.
This is why AI hallucination is such a huge problem. Therefore, in its stupidity, AI can’t discern concepts like people, race and stereotypes — “their fealty to superficial statistics drives this horrific stereotyping” the author says.
Hofmann’s summation below puts things into perspective:
If we don’t know or understand how LLMs are being used in areas like housing/mortgage decisions, business loan decisions, crime proceedings, healthcare and social programs, politics, access to VC capital and much more. What we don’t know can hurt us and so too LLMs will harm humanity and set us back generations.
Since the horrific George Floyd event, many in the Black community have wrongly delved into the emotional abyss of Wokeness, with a slew of nonprofit organizations brought to the forefront as drivers of entrepreneurship. This is upside down — nonprofit organizations leading profit venture/entrepreneurship are simply counterintuitive, counterproductive, a huge distraction and plain stupid.
Everyone knows that the world comes down to power and money, and if you have neither you may suffer if trying to attain power. So once again Black peoples’ future prosperity rests in authentic entrepreneurship creation and particularly in technology because that’s where the wealth and power reside. So if we are not at the power table doing the engineering and making the executive decisions we stand no real chance at power.
As Marcus Garvey once said intelligence rules the world and ignorance carries the burden, so unless we focus and apply intelligence to our aspirations, we’ll forever carry the burden of the capitalist system for the economic benefit of others. We’ll increasingly become the useful suckers of the system, existing in a vicious circle of pain, suffering, and despair. Ever distracted in an infinite state of being lost, perplexed, and confused.
Don’t wait for others to make change for you, no one will share their gains with you, ill-gotten or not. No one is coming to save you. Absolutely no one! We must rescue ourselves. It’s up to us to be the arbiters of our own lives. This is how the real world works — it’s not a morality play it’s a power play.
Personal responsibility, self-reliance, and determination, with empathy, underpinned by courage and a willingness to venture away from the herd. Finding our individuality and entrepreneurial drive gives us the best shot at succeeding during our twinkle of existence in the universe.
The LLM challenge is just another challenge to overcome and as they say, you can only fight fire with fire. So fight covert LLM racism by applying your intelligence technology to the pursuit of tech entrepreneurship, overcome fear and go with courage. There is no reason why we can’t succeed if we go for it intelligently, with persistence and determination. Equity in tech is not an option it’s a must. We must become owners and developers of tech and not just consumers of it.
