So, are you among those who believe that AI or Artificial Intelligence is intelligent enough? If yes, be ready to get your known facts turned false in this article. The former professor and an avid researcher Roger Schank once came up with a proposal related to AI. He stated that a computer being the best example of AI should be capable of watching the West Side Story and also recognize the actual plot of the famous Shakespearean drama Romeo and Juliet. He along with his team of students relied on the fact that the stories are completely central to reasoning, intelligence and the moral. But the result Schank came up with is quite shocking. According to him, AI is at all intelligent to the extent people consider.
Feature On AI
An article published by HBR.org on Artificial Intelligence is looked upon as an ironically reliable example of the type of work that the computers are incapable of doing yet. Two experts after carrying out years of research using their experience formulated a thesis by assembling all the evidence and come up with this article. Before you go ahead reading a short review on this, remember that neither the fact that a software is incapable of writing an article is something to demean it nor it is something that won’t ever be transformative. Instead, it is just an image that shows the exact manner in which machine learning technologies or AI functions, their pros and cons and how can you develop them into good writing tools for use in the future.
Predictions- They Are Annoying Enough
Today, Artificial Intelligence functions by simply formulating the tasks as some prediction issues and then straightaway using the statistical measures and enough quantum of data for making some more predictions. Auto-complete or auto typing is one such prediction issue that is text-based. And there is hardly anyone who hasn’t yet experienced such an issue while typing a text on the mobile or computer. When a research work was conducted, a lady stated that whenever she began typing “I” in any text message, her phone comes up with some suggestions like “was”, “will”, “am” etc. Now how does this happen? As stated before, your phone is relying on the statistical and data modeling for predicting the next thing coming up.
Now, this is not a fact worth appreciating. The lady continued saying that once she choose the word “was”, the phone came up with the next text prediction- “working”, “sleeping”, “eat”… Once a word is chosen, without giving any space it moves on with its next prediction and finally ends up causing a typing error. No no, it is not that she entered a wrong text. But the phone ends up thinking that your entered message is a mistake you made on your part and comes up to solve it by using its prediction- which possesses the capability of turning the actual meaning of your message into completely a different one.
Prediction issues in case of machine learning are termed as ‘supervised learning’. Now while this is helpful when you are a bit confused with the spelling of some specific word, it is really annoying when you are cautiously typing a text and some words are automatically getting altered as a result of the predictions being made by the phone.
However, it is not the prediction problem which is why AI is said incapable of writing a blog or magazine feature. In a recent conference conducted on Artificial Intelligence & Journalism, Sam Bowman, a professor serving at the New York University stated that giving shape to the concept of generating long coherent texts irrespective of the same reflecting a very clear meaning is something beyond expectation at least within the next few years. Researchers have revealed that machine learning is capable of generating coherent texts only in certain specific settings. However, building real-time systems capable of going all way from a certain abstract idea into a complete coherent text is something that still seems difficult.
For illustrating this difficulty, bowman adhered to a screenplay which was titles Sunspring and was written with the help of machine learning. This script was finally generated by adhering to few dozens of Sci-Fi screenplays into one neutral network which means that the specific units of data which the algorithm was then learning from were nothing but a single text or character.
Given all the characters which came beforehand, the algorithm was using the same to make a prediction of the characters which can come up next. And the result? Well, it’s enough to make you dye laughing with the sense they generated. And thus Sunspring written using AI never managed to display itself as a complete story.
Artificial Intelligence Generated Summaries
Summaries are one section of writing whereby machine learning has already shown some progress. Analysing and coming up with the essential parts of a particular text and helping with the production of a summary is looked upon as one of the common most writing tasks. Talking of the Press or Journalist teams, they compile certain clips together to form a complete news package for the day while reporters at the time of writing stories summarize prior developments, the editors summarize chapters and so. Now some part of this work can be done using machines and the tech firms and startups are seemingly in a race to come up with products and tools that can help in making it highly accessible.
Talking about the auto-summarization method, it usually fits into either of these categories- abstractive or extractive. The extractive method attempts to identify the important sentences from a bunch of few sentences and then come up with the creation of a summary by simply placing them one after the other. However, the modern versions related to this technique are a bit complicated but the actual idea which was introduced by Hans Peter Luhn in 1985 in IBM offers a concrete sense towards the approach. According to him, words which are most used in any document are the ones which offer some concrete clues to the subject of the document. Hence the sentences containing these commonly used words are looked upon as the most representative of such documents- and by extracting these sentences and getting them combined into one single paragraph, you can head with the creation of a concrete summary.
The Future of AI & Writing
When AI is there for over years, why didn’t the same proof capable in the field of writing even in the imperfect form? One reason judging the same is the culture. However many writers consider there is no need for such kind of tools. Also, computer scientists never seemed concerned with the fact about how people are going to appreciate and favor their work. In the words of Ani Nenkova, in case of auto-summarization, the main focus remains on accuracy improvement instead of thinking regarding how this technology can be used in a particular tool that people can put to use in reality.
Another essential factor is money which most of the newsrooms and writers lack. Finance is a major area whereby machine learning has made a great impact because money in large portion has been there for giving it shape. And the ultimate reason that these tools couldn’t make enough of a dent in case of writing is just that whatever results have been extracted didn’t prove good enough for serving the readers consistently.
Enough though there are drawbacks when it comes to writing using AI, it cannot be denied that now technology is undergoing numerous changes and you can surely expect to get in hand something far better in the next few years.
Norjimm, a leading web application development company India offering all kinds of web app and mobile app development solutions at affordable prices. Our highly efficient and experienced engineers possess hands-on experience in huge and complex application development projects by creating robust and scalable applications across latest technologies that make us stand out in the crowd.