AI.. What do we want?

In fairness, this is a rather wide and open-ended question. One that requires a lot of thought. It is dependent upon what one perceives as AI, therefore, expectations will differ. Personally? I believe that communication is the first key to successful integration of AI into the human psyche and herein lies a problem.  As I have said previously, there are many systems available that can answer questions, some of them quite complex, but they lack the element of empathy. The questions must be worded in a particular way and the responses are somewhat automated.

 

brain.png

The image above is not our brain, so we don’t respond as such. AI is a computer, as it stands, and that is the issue. Incompatibility, especially when communication is necessary. The preprogrammed response that is currently the norm will no longer suffice if AI is to become a widely accepted part of everyday life, but what is the alternative?

A group of researchers, led by Professor Komatani, at the Osaka University are developing a protocol that involves implicit confirmation. Much of the issue surrounding AI understanding is that is cannot confirm the meaning of a new word in its lexicon and requires explanation. That is okay in a scientific environment, where the word can be ‘explained’ but not in a domestic setting or in an autonomous vehicle, for example. The computer can not learn, nor can it guess if it doesn’t have the tools. The implicit learning system allows the AI to use the word and check whether it gets the predicted outcome. If so it can add the words or phrases to given scenarios.

The system can be split into two distinct forms:

Machine learning – Using algorithms to analyse the data and learn, then using the data to make predictions. It includes both supervised and unsupervised learning processes, followed by reinforcement from responses.

Dialogue learning – AI learns to communicate and converse in a natural manner through implicit confirmation.

The above is somewhat simplified but does show where AI can and will go in the near future. But is it enough?

For now, it is another step forward and our personal assistants will sound a little less like automatons, but the key is in intelligence, and how we perceive it. We process so many levels of information simultaneously, from a full set of sensors. AI cannot do that yet and it is doubtful that it will be able to in the near future, but research is getting there, and new technologies are presented on a daily basis.

It is a long way down the AI information highway and I think it is fair to say that we are just approaching junction one.

 

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s