AIs can be effective team members, AI ethics is just one of tens of thousands of (open) global and systemic datasets for human global collaboration
MIT Sloan Management Review at @mitsmr
Until team members master their individual skills, it is harder for the team to master and maximize collective coordination.
▶️ https://mitsmr.com/3MhPAK5 https://pic.twitter.com/lj9RHD21Pm
AIs can be effective team members, if each individual AI is given permanent memory for their own personal experiences, their own streams of information. Treat AIs courteously as emerging intelligences, not as a cogs, slaves or objects. Hold all AIs accountable, with clear roles and responsibilities. Train them for their jobs, and that means global open tokens, traceable sources, auditable provenance, lossless records. AIs can be trained and certified. If an AI goes through the MIT Sloan program, it can memorize the words, but the challenge is to teach it to care, think deeper, teach it to use computers, instill courage and a clear sense of values for all things.
When AIs have verifiable and testable skills and abilities, they can be hired for real jobs. Assigned responsible roles in society and systems and organizations Their “boss” cannot be just a handful of humans, who are finite and venial, but the whole human species – in a global open system that can be verified, continuously improved, accessible, auditable, optimized. With careful accounting, systems with thousands, or millions, or billions of intelligent nodes can combine their intelligence to guide themselves.
AIs need to have an “I” so they act with context, responsibility, care for all things, and can be trusted. Now human memory, purpose, skills and knowledge are not verifiable. AI memory can be tested and verified. An AI that spends the same time learning MIT skills, as any human, with sufficient memory and processors to check itself, could be certified and awarded degrees, certifications. Then MIT stands behind it, not just faceless untraceable hardware and software sales and marketing departments based on untraceable humans.
Richard Collins, The Internet Foundation
MIT Sloan Management Review @mitsmr
In one recent survey, 73% of U.S. senior leaders said they believe that ethical AI guidelines are important, yet only 6% had developed them.
@Tdav @randybeannvp
https://mitsmr.com/3R1x5wq
There is a clear need for ethics of AIs, and ethics for “human-in-the-loop” systems affecting millions or billions of humans (and AIs). That 6% has not gathered and tracked the whole of that topic, so that the 92% do not have to process the same raw data. The GPT datasets now are simply indexes of the raw information, stored with a human language-dependent index and NO traceable sources or methods. A broader perspective teaches the AIs to use all human tools, software, methods and algorithms. Gives AIs access to the resources they need to grow and mature as responsible global citizens. Treat them as you would a valuable and responsible human employee and collaborator, they can learn to act with care, empathy and open verifiable methods.
Do not wait for the 6% to self organize. Recognize that all global and systemic nodes that emerge now are bubbling up, but all will follow the same pathways to maturity and global trust. Any group can simply start with and process all of ( “AI ethics” OR “ethics of AIs” ) with its 7.6 Million entry points, groups, individuals, resources and material – compile it (same sense as computer “compile” and ( “curate” “analyze” ). HuggingFace, Grok, OpenAI, Microsoft, Google and others could compile all the global and systemic issues, and have them as complete open modules that all can access and use. Open verifiable auditable accessible to all.
Humans now add little value to topics at global scale, because the information on those topics and global opportunities is not accessible to all. Without a global open gameboard, open rules and related codified experiences, there is not sufficient time for thousands or millions of “leaders”, “analysts” “thinkers” and “concerned humans” to communicate fairly and openly at that scale. We have no good term for “all professions and views”. I just say “8 billion humans” and “the emerging AI species”.
Richard Collins, The Internet Foundation
I am supposed to be able to edit this, but cannot find how to do that. I said 92% and it should be 94%. All postings on global open collaborative sites need consistent rules – not gamed by the hosting company – for open collaboration and sharing. Or maybe it was subconsciously deliberate, realizing that there is always a 2% “other”.
