Principal Analyst Data Governance | Posting commentary for analysts since 2017 | Brier Score of 0.211 | Experimental science: show me the evidence | Veritas filia temporis | Views mine own
This is a useful view into the #airisk surface. The malfeasance occurs where there are insufficient controls and tests to distinguish a human knowledge agent from an artificial one. The story goes to the importance of #identity as a public good: ordinarily we think of identity about the individual but here it is direct financial loss to the organisation. What's more there is crowding out of human talent, a dent in the #innovation curve and yet more opportunity cost. Yes, this is the race to the bottom and a future akin to the film Idiocracy. Note these scams do have remedy as they are variants on mail fraud, but that won't return the money nor address all the above costs. If now already, folk should turn their minds to reconsidering how to quality assure their funding validation and certification processes
All of our security and verification systems are built around very specific intuitions about things humans can do that machines cannot (& we are still fooled lot). We are very unready for how LLMs will overwhelm these systems. This example of AIs being enrolled in college courses to get grant money is one of many examples we will soon see. https://lnkd.in/e_bT35_U
