Recent news articles are in-depth in discussing the latest advancements in chatbot technology- Meena: Google's new chatbot. According to Google, this is going to be the finest of all chatbots that humankind has witnessed. This is because of the fantastic inbuilt features of Meena that have been stated in further more details in Google's recent research paper. Also, Google claims that this is going to be the best voice assistant to date. Although it hasn't been launched yet, the world is looking up to it.
What can Meena do?
Meena is a multi-turn, free domain chatbot with about 2.6 billion parameters that facilitated advanced training involving a capacity to digest 341 gigabytes of web text. This training process lasted for 30 days. This vast training dataset has prepared Meena well for conversations about anything on this earth, as Google claimed. It is more like a human and a versatile chatbot. Each conversation is turned into a pair, namely context and response team, with which most relevant answers are given. Following is an instance showing a conversation between a human (volunteer appointed by Google) and Meena.
Have you ever seen a bot cracking jokes? Here, Meena does this. Because of this unique feature, Google has tried to create an actual companion. Here's a small snippet to showcase Meena's joke-cracking talent.
Meena is well equipped with techniques enabling it to give relevant responses. In the above two snippets, there is no sight of Meena giving irrelevant answers because of which no one can make out whether that person is talking to a bot or a real human. This can be well explained with the help of a graph.
In the above graph, Meena's performance has been compared to various other, previously developed chatbots plotted on the x-axis concerning the human-like conversation feature plotted on the y-axis. This human-like conversation feature can be measured with the sensibleness and specificity average or SSA, which considers the sensibilities, relevance, specificity of the answers given by the chatbots. The above gives us a relatively clear picture of how advanced Meena is in her responses. It is quite interesting to note that these kinds of progressive reactions have been achieved with no such different and modern techniques, but existing methods have sufficed.
There have been many instances wherein chatbots crash due to unexpected or very lengthy input. Still, fortunately, Meena handles errors effectively and similarly. A human would simply say that it doesn't know and as far as the crashing of chatbots is concerned, it has advanced, enhanced capabilities of processing large amounts of data at a time. So, these advancements have coped up with all significant shortcomings of the previously developed chatbots.
In conclusion, there are furthermore advancements required by Meena to handle conversations involving practical task's discussion like
In the above conversation, a practical outlook has been presented, which emphasizes on Meena's shortcomings.
At the end of the day, Meena remains to be one of the finest chatbots ever.