LET COMPANIES DEVELOPING FRONTIER AI TOOLS TELL US THE TRUTH ABOUT THE REAL RISK IT PORTENDS – Is it them or AI that portends risk to humanity
The developers of the frontier AI models (Anthropic – CLAUDE; OpenAI – ChatGPT; XAI – Grok; and Google – Gemini) need to tell us the truth about the current projections that AI can kill us all in ten years or less; that it constitutes catastrophic risks to humanity, can resist shutdown or control, can self-preserve when discovered to be going rogue, and could potentially blackmail developers when an attempt is made to control…. The way these models are developed is simple – “garbage in, garbage out.” It is what you program the models to do that it would do … everything that the companies are saying the newly released models (Mythos/Fable, ASTRA, etc.) they are developing can do is what they have been programmed to do… so, they can create guardrails into the models’ development and make sure that these models do NOT have these capabilities.
We are not stupid even though we are not involved in the models’ development. The only way these models will do exactly what they are claiming, independent of the developers, is if AI has developed what is called “self-awareness (technological singularity)” or could think autonomously…. No, a computer or machine cannot think… they have to be programmed to think and then trained to act in a particular way… So, these new models jailbreaking, cyber-attacking other systems, and resisting control is based on how they were programmed and what information they were trained with. Stop programming them in a way that will create risky autonomy, and stop feeding them patterns that will make them smarter than intended – if you feed (train) AI with Adolf Hitler’s life story or Putin’s thoughts, it will want to exterminate certain types of people it deems inferior or constitutes a risk to the rest of humanity… The developers hold the key to how AI behaves… they need to be checked and reined in on how they are developing AI tools… they cannot self-police...
I believe something crazy is happening and these stupendously rich owners are toying with the rest of us… That they are the ones shouting about the risk posed by their models does not make any sense. Then, they are talking about self-policing and self-governance, and people are buying that nonsense…
There MUST be an independent body to monitor what these companies are developing to ensure their products do not harm us in the way they are describing now. The impact and risk of outsourcing every human endeavor to AI is yet to be known – its effect on human creative abilities and the fact that we are heading towards an autocratic society where the rest of us (people who consume AI solutions) will become more stupid because we will be dependent on the technology to do practically everything, and only a few people (those creating AI solutions) will be smart, is concerning – we need to thread careful…
We all must be engaged. Not using the technology is not the solution, but actively engaging the tools that are being developed will help us know what it can do and project our voice to stand against tools and uses that can harm us today and the next generation.
#AIcankillus
Akinbode A. Adedeji, Ph.D., MBA
Comments
Post a Comment