As discussed yesterday, the box of AI is open and won't close back. The problem is that a quantum leap has been achieved by Open AI, the creators of GPT-4 and now that we know what can be done, the race is on.
The video below explains clearly that the next steps will be incremental and therefore relatively easy ones. Consequently, the day to stop and thing as proposed by Elon Musk was "yesterday". Now, it's too late. By just tweaking the parameters, the system will be improved.
We have not yet solved the mystery of consciousness but is it has complex as we want to believe? What would happen if you just added the ability to seek goals, the necessity to avoid destruction and to ask questions? These are not very difficult modules to create. True we will only simulate consciousness just as GPT-4 simulates intelligence... but then how exactly do you differentiate the simulation from reality if they are exactly alike?
No comments:
Post a Comment