Sallam Abualhaija is originally from Jordan, where she completed her bachelor’s and master’s degree in computer sciences after which she went to Germany to do her PhD at the Hamburg University of Technology. She talked to Silicon Luxembourg about natural language processing and the misconceptions and future of AI.
The focus of her PhD is related to natural language processing and bio inspired optimization techniques – finding a correct meaning of a word in a context, using some advanced algorithms, which are closely related to AI.
During her post-doctoral, Sallam worked in crisis management – specifically social media mining and rumor detection during crises, in collaboration with relief organizations. Now she focuses on applying AI for solving different software engineering problems.
“Among the projects we work on is checking compliance of legal documents according to the requirements of GDPR, and we also do ambiguity detection, and providing text interpretation across sectors“, she explains.
What Is AI?
When hearing the abbreviation AI, many people do not know exactly what that means and encompasses.
The image, she says, of robots doing things instead of people (and looking like people), is not really precise. It is not near to the truth – machines can converse, drive, analyze data, drive, make decisions – things that, for a long time, we thought were exclusively reserved for humans, and it turns out – they are not.
“AI approaches those tasks and tries to solve them to the best of their ability in a human-like way, in a way mimicking what a human would do,“ she explains, emphasizing that you must remember to exclude emotions.
But, machines do try to rationalize the way humans do.
“One of the good examples is a conversational bot. It does what it has to do, going through steps which it learned, but if it encounters problems there will be a human assistant that will take over,” she mentiones.
The point of the entire conversation is to show that “the AI is not somewhere in the future, it is very much in the present, very much a part of today – face recognition, translators – take a look just how google translate changed within the years. Then there are conversational bots, like Siri and Alexa, event extraction from emails, recommendation of similar videos, ads, etc. Without being aware we are interacting with a lot of AI technology,” Abualhaija explains, talking also about her own research, which is natural language processing and machine learning.
“[…] let AI take those jobs in the next generations, and we will be training young ones to think and innovate.“
Sallam Abualhaija
Natural Language Processing
Natural language processing is a sub-field that enables machines to create a sort of understanding of the textual documents which we focus on. Then it makes decisions. Machine learning is „training“ the machines so they can notice and then predict different patterns in the text.
While doing the compliance checking of textual documents from the legal domain, they try to teach the machine about key words, word clusters, how many modal verbs are there, and then based on the features that we learned and understood machine learning provides with a clasifier that predicts on text documents.
“Machine learning is very interesting, sometimes we just provide the data to the machine, and it learns about, e.g., the frequency and cooccurrences of words. Usually we do not „feed“ the text as-is to the machine. Instead, we create some sort of mathematical representation of the text and give it to the machine to make sense of it. The machine then looks for patterns over numerical data. For instance, we provide the representation of some legal document, and the machine needs to know whether it complies or not compared to the patterns of already seen documents,” she explains about machine learning.
Scientists feed it data, and the machine spits out a conclusion.
The Future Of AI
“This is deduction and intelligence,“ says Sallam, mentioning pop-cultural references about The Matrix and The Terminator, and the fears that many have when it comes to AI.
“We humans are evolving continuously, and these rapid leaps of technology are changing our lives. There is always room for danger, not only in this field. And danger usually comes from how humans use this technology. AI is abused, yes, take a look at deepfake. It can change the direction of a community, how it is perceiving something, and it is serious,” she explains, admitting that she also thinks about a future where robots are replacing humans in one way or another.
Things that we have seen in science fiction as children – are a reality now, and while it is normal to be afraid of what is new, it is also in human nature to adapt – there is no other way but forward.
“AI will wipe out a large portion of workplaces – and that is totally fine. Then humans will be able to quit those routine jobs that require a lot of effort, and time and we humans, as intelligent creatures, will use our energy on more innovative tasks. So let AI take those jobs in the next generations, and we will be training young ones to think and innovate“, concludes Sallam Abualhaija, pointing out that AI will also create completely new jobs.
