AI, sentient or not, we need practical applications
There have been too many discussions on whether chatbot LaMDA is sentient, while
our society is in urgent need for well-designed/practical AI powered products, to help people’s every day life.
Our senior citizens could use “conversation companions”, especially for those with early stage dementia and live alone. This companion should have the abilities:
- to initiate engaging conversations with the target person, not just responding to person’s question, know how to stop and change subject after detecting person’s negative responses or boredom.
- to recount old-time stories that the person is familiar with/interested in, such as a memorable NBA game, that would exercise the person’s brain cells, keep the person mentally active, and maybe slow down the progress of dementia.
- to remind the person the time to take medicine, meal, or nap, and get verbal confirmation from the person that this is done, and play back when the person asks again, “did I take the pill after lunch?”.
- to announce the person’s children/grand-children’s social media postings, read out the contents or display pictures or videos as they coming in, to fill the day with joy even living alone.
- to send alerts to pre-configured agency or relatives when it detected emergency through its audio and video sensors.
Our children could use advanced AI chatbot (or virtual tutor) to learn. From birth to age 3, scientists see the fastest rate of brain development in the entire human life span. An AI powered “virtual tutor” can be additional stimulants in this critical stage, or a desperately needed early child educational tool for disadvantaged families that lack the resources to get quality child cares. The virtual tutor should be able:
- to adjust the speech to use Infant Directed Speech (high-pitched, vowel-stretched and emotionally charged) with easy vocabulary when talking to a baby
- to listen a child’s uttering, make corrections, or help a child to finish the sentences - learning has to be 2-way communications.
- to assess a child’s development levels along the way, show results to parents.
- to allow parents to decide how a child should grow up, by adjusting the chatbot when conversing with a teen-age child, in formal/cultivated languages, with chosen accent based on geological or cultural background, or in vernaculars mixed in with Gen Z slangs to make your child popular in school.
I have gone through the (only available) LaMDA conversation multiple times, the questions are very clever, and responses are very impressive. But with limited information, I don’t see it can do all of above, especially in customization area: blend in personalized data for individual’s hobby and interest, medicine schedule, might as well include data of medicine’s side effect for AI chatbot to pay attentions to during conversations. And the challenges of adjusting chatbot’s voice, accents, use of words and sentences - easy or sophisticated, formal or colloquial conversation.
Hopefully there are products underway that would provide some practical benefits to average people, and looking forward to these applications.