0
$\begingroup$

Just a thought question that came into mind from the current examples in the world. Any AI solution created so far for making the human effort lesser also has a tendency of no human effort at all. Are there any efforts on this aspect of the symbiotic between human and AI ?

Aspects like how to

  1. Contain ? - Like humans with idendity
  2. Control ? - With Human decisions and Human collaboration and Human consensus
  3. Maintain ? - Learning life-cycle with humans

Are there any best standards or frameworks for AI Governance which is agnostic of any AI and can manage the above aspects separately ?

An example below of this symbiotic:

An example blockchain system of human collaboration and consensus tied to AI learning life-cycle and control aspects. The more learning the more collaboration and consensus of humans will be involved. In this example AI will be like a fitness trainer helping Humans to collaborate and work better and aimed towards truly augmenting the green circle in the diagram.

enter image description here

Still the question remains unanswered is that should AI be treated as a external entity and should there be any clear protocol enforced (principals / laws / governance) for any interaction with human and their resources ?

Should every AI identify itself as a Human extension with a unbreakable link to a Human before handling human owned resources so that the responsibility chain for human augmentation is established ?

$\endgroup$

put on hold as too broad by nbro, respectful, malioboro, ashenoy, John Doucette Nov 19 at 0:53

Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • 2
    $\begingroup$ Your question is kind of unclear. AI do take human help, for example in ML human labelled data is used. In RL AI can learn from human actions. So what exactly do you mean by symbiotic relationship? $\endgroup$ – DuttaA Oct 4 at 13:49
  • 1
    $\begingroup$ The question in short is about the standard protocol that ensures and guarantees that AI will need Humans for any of its transaction. The protocol doesn’t reside in the ML library it is external and audit-able. The protocol that ensures that the goal of AI is aligned and tied up with the goals of Human consensus and Human responsibility of resources. $\endgroup$ – Jeetu Oct 4 at 15:11
  • $\begingroup$ There are certain aspects of the human faculty, namely creativity, speculated to be un-automatable. Note the term "speculated." $\endgroup$ – Tautological Revelations Oct 11 at 18:41
  • $\begingroup$ Human creativity will be outrun some point when speculations won’t even apply to AI. Ex: A car helps human by augmenting motoring. An AI car also helps human by self-driving and reaching destination safely and associating itself to its owner so that the responsibility for any accidents is established. The decision making is aligned by humans (on destination) on human owned resources (car, road, insurance) and extended / regulated / identified to AI (on safety). It is important to ensure this human regulation in any AI. $\endgroup$ – Jeetu Oct 12 at 15:34
0
$\begingroup$

Training data, we can continue to be observed and understood, I imagine something trying to constantly increase would want to keep humanity around so it can better understand the species that brought about its existence

$\endgroup$
  • $\begingroup$ The idea was about establishing a clear distinction and audit-able protocol for interaction with human and their resources. Example USB protocol that every PC understands. Need a similar one that every AI will understand and follow ? In the diagram I have depicted the resources in the human side as there is responsibility. AI cant be held responsible or accountable for anything if I am right ? So again the protocol that depicts the chain of responsibility tracing back to humans which AI will need for any transaction $\endgroup$ – Jeetu Oct 4 at 14:26
  • $\begingroup$ I get what you mean, something we can impose to ensure it will always need us, in that case i do not believe anything applies, eventually general ai will come, then sentience, and after that, when ai can write code itself, control machines itself, there isnt anything we as humans could do for it better than it could for itself, with (as i mentioned) the exception of us being ourselves and existing for it to observe as species of intelligent beings. $\endgroup$ – nickw Oct 4 at 14:35
  • $\begingroup$ Not until AI has unlimited access to resources. The protocol should make this clear distinction of resources and its guidelines. This is why in the example I have mentioned AI transactions. If every AI transaction on a human owned resource can be audited externally and traced back to a human responsibility using some protocol we might get a accountable AI with more human engagement on resource control. So maybe an open ledger of transactions with human consensus involved. A protocol where every AI will learn human principals to gain human credibility. $\endgroup$ – Jeetu Oct 4 at 15:11
  • $\begingroup$ I'm not saying that couldnt be done im saying at a certain point, since you question says "forever", it will be able to write any protocol out of its own code, on the timescale of forever i dont see why an ai would "need" anything from us after reaching a point it can self improve. if "not until ai has unlimited access to resources" is a constraint you should reword the question. $\endgroup$ – nickw Oct 4 at 15:31
  • $\begingroup$ Is it right to think that it is a Human written protocol that any AI should follow forever and not the other way. It is the Human written protocol or standard that determines the guidelines for consumption of a human owned resource and any change made should be made by an accountable entity - Human $\endgroup$ – Jeetu Oct 4 at 15:55
0
$\begingroup$

The communication protocol between humans and Artificial Intelligence systems is surprisingly easy. A human manager has to communicate to it's own group what the AI wants. Very similar to a priest who explains to the tribe what the god's desire are.[1] It depends on the human manager how well he plays his social role.

To investigate which development was made in the past, it make sense to analyze existing communication structure in the society. Which kind of spokesperson is already available who negotiates the needs of an AI character to a human group? Are any AI experts available, who tell to the ordinary people how a self-driving car is working internally, and how to build intelligent machines? It's a rhetorical question because the answer is known already. In today's society, the priest of the technology revolution are located in the media and in the technical university. They are building the infrastructure to communicate between the needs of an AI (which are built already, or which are built in the future) to the ordinary people.

The standard protocol what Artificial Intelligence is, and how to interact with it, is provided by mass media and their participants. This kind of centralized communication protocol was modified in the past with the upraising of the internet, but in general the old structure of experts located in the media and in the university sector remains in power.

[1] Postman, Neil. Technopoly: The surrender of culture to technology. Vintage, 2011.

$\endgroup$
  • $\begingroup$ My question is about how to contain ? control ? maintain ? AI as a true augmentation to a Human like an extension of a Human. In the priest example – there is (no contain !) and (no control !) without both I doubt a symbiotic can be maintained between AI and Human Why does AI need a Human Manager to teach it when it can teach itself. Human can’t teach AI forever $\endgroup$ – Jeetu Oct 4 at 21:35
  • $\begingroup$ I was seeking ways for a truly augmenting AI maybe (1) Decouple AI from all Human owned resources, decentralize AI and associate it to a Humans identity for chain of responsibility – Contain ? (2) Humans work together in signing off an insight driven AI transaction from multiple insight options provided by multiple AI – Control ? (3) Record credibility of the transaction on a open ledger in the Ethereum – Maintain ? $\endgroup$ – Jeetu Oct 4 at 21:35

Not the answer you're looking for? Browse other questions tagged or ask your own question.