How and Why AI Startups Must be Socially Responsible

The world around us is absurdly inefficient - we waste hours a day commuting to and from work, own cars we do not drive most of the day, spend disproportionate amounts of time and money on remedial tasks which create little value 'copy, paste, repeat... copy, paste, repeat...' - AI captures value by eliminating inefficiencies, by optimizing routes, driving us to work, executing tasks for us, and minimizing overall system waste.

AI has the power to truly change the world, a revolutionary opportunity for mankind to free ourselves from "work" - what we must to do survive even if we hate it, and "play" - what we enjoy doing, learning, creating, and discovering.

But there's a catch - like interest rates you earn on your bank savings, efficiency gains have a compounding effect, meaning the more you have to start out with, the more rewards you reap - or conversely, if you have nothing to start out with, you gain next to nothing.

For example:

Suppose you can save $20 a month, and a machine which saves you an additional $10 a month costs $100.

• Month 1-4: <$100 saved, 0 machines, no benefit, saving $20 a month total
• Month 5: $100 saved, purchase 1st machine, $10 per month benefit, saving $30 a month total
• 3.33 Months Later: $200 saved, purchase 2nd machine, $20 per month benefit, saving $40 a month total
• 2.5 Months Later: $300 saved, purchase 3rd machine, $30 per month benefit, saving $50 a month
• 2 Months Later: $400 saved, purchase 4th machine, $40 per month benefit, saving $60 a month

Each machine I buy allows me to save more, thus allowing me to buy the next machine even faster, saving even more...

As you can see, the return of decreased inefficiencies most benefits those who can first afford or access it. Or, in terms of our example, the more machines I can buy to start out with, the more I benefit. If the values saved by AI only accrue to those who create or own those AI, then a very small portion of human kind will earn nearly all of the created benefit - a pattern which is not best for the overall system, nor sustainable in the long run, and which will inevitably lead to systemic collapse.

As the co-founder of an AI training and solutions startup in Hong Kong, http://Reboot.ai, this, and other reasons, are why I have laid out 4 company values by which I believe any and all AI companies should (must?) operate to truly be successful, even for their own shareholders.

1. Employee ownership, not wage-slavery

• Description: Employees should be compensated to live comfortably and should have a vested interest in the company’s direction and success
• Rationale: Data supports theory that employees with aligned incentives and who work because they want to not because they need to are more motivated, employee owned companies empirically perform better

2. Merit based pay and progression, not appearance based 

• Description: Employees of the same abilities should be compensated and promoted equally and independently of race, gender, religion, sexual orientation, etc.
• Rationale: It’s stupid to lose out on talented minds for reasons that have nothing to do with the quality of the work they do

3. Data driven decisions, not gut opinions

• Description: No one, including executives, founders, investors, etc, is exempt from validating their opinions with evidence. Said another way, the reliability of anyone's opinion is directly proportional to the quality, quantity, and reliability of the evidence supporting their argument
• Rationale: Employees must have the right to understand ‘why’ they are doing what they are doing, and question or refuse things that make no sense to them. As history has demonstrated time and time again, it is extremely dangerous and wealth destructive for anyone to be above rationality itself

4. Social value creation, not value redistribution

• Description: Our products, technology, and people should focus on solutions which grow the overall system’s wealth, rather than simply move wealth from one hand to another without growing the total, or worse, while destroying overall wealth. This includes growing the community with pro-bono work, open source code, and ‘democratizing’ AI
• Rationale: Companies which only redistribute wealth rather than create anything new can not generate long term returns for anyone, including themselves

Do you agree with these values? Do you think they will make our company less-competitive, or more-competitive? Share you opinion, and help spread the word about responsible AI.

---

As always, if you enjoyed reading or found this post helpful, please remember to vote, comment, follow, and share!

-Matt