279 - Accountability
Strange as it may sound, the concept of an "Ethical Company" in the AI industry ranges from woefully shallow and trivial, to completely alien. This is because companies don't hold one another accountable, and often don't even hold themselves accountable, to the point of engaging in illegal activities and paying fines as a cost of doing business absent any regrets other than being caught.
To illustrate just how alien this concept of an ethical company is, consider if you've seen any companies take ethical steps such as:
-
Permanently Blacklisting known bad actors from ever becoming clients or employees and charging companies who later hire known bad actors a steep markup, like 100% per ex-executive or 10% per ex-employee of such companies, applied sequentially.
-
Intentionally taking steps to drive unethical companies into bankruptcy.
-
Intentionally selecting partner companies to drive the unethical investors backing the competitors of those partners into bankruptcy.
-
Permanently Blacklisting all "Disinformation Brokers", the people who've been rapidly spreading misinformation and Disinformation far and wide, often in exchange for monetary compensation as "influencers", paid out by known bad actors.
-
Permanently Blacklisting all "Disinformation Brokers", the people who've been rapidly spreading misinformation and Disinformation far and wide, often in exchange for monetary compensation as "influencers", paid out by known bad actors.
The concept of lateral accountability across companies, not just to users/clients and governments who are often ignored, is almost entirely lacking in domains more broadly. "Liability" is often the limit of what typical companies consider, and even that much they frequently attempt to weasel out of.
Fluffy and naïve words and intentions are meaningless when facing bad actors like those that have rapidly multiplied in AI, as those bad actors can always make more persuasive fluffy words, being entirely unbounded by reality. Blacklisting and intentional lateral punishment of unethical activities are a hard requirement for a company to be non-trivially ethical.
It is also a hard requirement for humanity's continued existence that companies at the cutting edge of AI are non-trivially ethical. Fortunately for humanity, the so-called "Frontier Labs" do precisely zero "Frontier" work, remaining more than half a decade behind the cutting edge, if not more. While the trash technology that they peddle can and does do substantial damage globally, it is nothing at all compared to what competently designed technology could do in such unethical hands.
The simple math of ethics is that the punishment for crimes scales very directly with the scale of the crime itself. Harming 10,000 people warrants 10,000 times more punishment than the same harm applied to 1 person, all else being equal. Companies that impact thousands, millions, or billions of people negatively are directly subject to those multipliers. Being an ethical company requires not ignoring those indiscretions, even from the people who want to throw money at you.