{"id":11792,"date":"2023-04-04T06:49:34","date_gmt":"2023-04-04T06:49:34","guid":{"rendered":"https:\/\/aurelis.org\/blog\/?p=11792"},"modified":"2023-04-04T16:41:01","modified_gmt":"2023-04-04T16:41:01","slug":"is-lisa-safe","status":"publish","type":"post","link":"https:\/\/aurelis.org\/blog\/artifical-intelligence\/is-lisa-safe","title":{"rendered":"Is Lisa Safe?"},"content":{"rendered":"\n

There are two directions of safety for complex A.I.-projects: general and particular. Lisa must forever conform to the highest standards in both.<\/h3>\n\n\n\n

Let\u2019s assume Lisa becomes the immense success that she deserves. Lisa can then help many people in many ways and for a very long time \u2014 a millennium to start with.<\/p><\/blockquote>\n\n\n\n

About Lisa<\/strong><\/p>\n\n\n\n

If not already done, please meet video-coach-bot Lisa<\/a> in development.<\/p>\n\n\n\n

But this text is also about any truly complex A.I.-project now and in the future. When talking about Lisa today, I also talk about the latter.<\/p>\n\n\n\n

This text is not an exhaustive expos\u00e9 or \u2018proof of safety.\u2019 It gives some ideas that I deem noteworthy for Lisa and the domain in general. Also, this text has an explicitly deontological take. The juridico-technical take is additionally important.<\/p>\n\n\n\n

Of course, we do our best to make Lisa safe.<\/strong><\/p>\n\n\n\n

Even that is not enough. The safety endeavor must include a continual effort to strive for optimal future <\/em>safety.<\/p>\n\n\n\n

That is not something to start thinking about when the stream gets rough. The stakes are too high.<\/p>\n\n\n\n

Particular safety<\/strong><\/p>\n\n\n\n

An individual user might encounter a unique issue. In any such case, the system must be prepared to react appropriately.<\/p>\n\n\n\n

No matter how many concrete rules we might put in place, this alone will never reach the desired safety level. People also don\u2019t proceed this way for themselves. We use rules of thumb in many familiar situations \u2015 mainly for speed and resource efficiency.<\/p>\n\n\n\n

On top of that, we also use broader mental pattern recognition in safety issues and more. The procedure is to recognize clear patterns of danger, safety, and uncertainty. When safe, we relax; we go into stress mode when in danger.<\/p>\n\n\n\n

The same applies to Lisa. The end goal is to play it safely and realistically.<\/p>\n\n\n\n

This is why an overly stochastic system doesn\u2019t pass muster \u2015 as also a system that rashly proceeds with uncertainty patterns.<\/p>\n\n\n\n

This means a hard job of human thinking, then of Lisa-aided human thinking. We don\u2019t need to wait until everything is thought through, but enough to play safe before proceeding.<\/p>\n\n\n\n

Particular safety is a lot of relevant work.<\/p>\n\n\n\n

General safety<\/strong><\/p>\n\n\n\n

This is not about very specific issues. Therefore, it may need less of a monk\u2019s work. It\u2019s a different job, but a lot of it anyway. This is, for instance, about the system\u2019s influence on sociocultural issues.<\/p>\n\n\n\n

One should not see such issues as collateral damage or things to be solved by society in the \u2018common domain\u2019 that would, therefore, be hardly relevant to the specific project \u2015 Lisa, in our case. <\/p>\n\n\n\n

Contrary to this, it belongs to the project\u2019s unmistakable responsibility.<\/p>\n\n\n\n

The \u2018solution\u2019 is not, therefore, a clearcut magical key. It is rather a meaningful contribution, an openness to share insights and a never-ending willingness to \u2018make this world a better place.\u2019<\/p>\n\n\n\n

This is not rule-based but intention-based. That doesn\u2019t make it less realistically relevant \u2015 quite the contrary. It does point to yet another essential change (of which there will be many in the A.I.-future): from rules and guilt to hands-on responsibility<\/a> \u2015 meaningfully based on the trust that people can be highly motivated by the latter. Fortunately, science robustly shows this to be the case IF people are not demotivated or ethically damaged<\/a>.<\/p>\n\n\n\n

With Lisa, we take this responsibility extremely seriously.<\/p>\n\n\n\n

Explainability<\/strong><\/p>\n\n\n\n

In short, explainability is about the system\u2019s ability to explain in commonly understandable terms why some decision has been made.<\/p>\n\n\n\n

On top of general and particular safety, explainability is also something we must take very seriously at every step. The user may ask the system\u2019s explanation for some specific decision, which must immediately be given or upgraded to developers. The human-thinking equivalent of the latter is when we run into some more challenging issue and need to take some time off to think about it, consult some other humans or books. The wise person knows when it\u2019s time to do so. We need to put such wisdom into the system, which can be done pattern-wise.<\/p>\n\n\n\n

The developers may also be asked for explanations for more general issues. Such explanations are to be viewed as part of the \u2018product.\u2019 They should be made publicly available.<\/p>\n\n\n\n

This does not entail opening up every detail of the internal workings \u2015 being commercially unrealistic. It does entail accountability of decisions, as one might ask of humans. No human knows the intricate details of personal decisions \u2015 far from that. Nevertheless, the concept of accountability is viable and effective.<\/p>\n\n\n\n

With explainability comes the possibility of co-creating safety by many users. Surely, this needs much common sense. It is one factor that becomes feasible in combination with others \u2015 then also important.<\/p>\n\n\n\n

In the very long term<\/strong><\/p>\n\n\n\n

Some people might mistrust super-A.I. by an unshakable principle. To be trustworthy, nevertheless, we put in place two more principles for Lisa safety also in the long term:<\/p>\n\n\n\n