Israeli startup Point.AI, located in the Sarona area of Tel Aviv, does not appear different from the other technology companies in the neighborhood. Like other companies with AI (artificial intelligence) in their name, it is a new startup that wants to use AI to revolutionize an entire market. Point.AI engages in data annotation for other companies developing AI algorithms and machine learning, but the company employs only high-functioning workers on the autistic scale. The idea is simple: to meet the demand for data annotation, while at the same time providing work for people on the spectrum.
Point.AI was founded six months ago by autism therapist and expert Tamar Dvir, CTO, business manager Tomer Gorovici, and technology veteran Eli Gorovici, who is no longer active with the company. "Together, we came up with the idea of trying to combine an existing AI need in the market, which requires a lot of data annotation, and a group of people who want to work and fit in, some of whom have capabilities that give them a special advantage in these jobs: good visual capabilities, close attention to detail, and the ability to work long hours at routine and monotonous tasks," Dvir tells "Globes." "In general, routine is characteristic that many people with autism need in order to succeed in day-to-day work."
The Israeli Society for Children and Adults with Autism (ALUT) says that there are 7,000 adults in Israel on the autistic spectrum. The organization estimates that 2,500 of them are in the middle or the high end of the scale, and can potentially work in technology. The organization adds that this number will increase as the rate of autism diagnosis continues to rise. Repetitive jobs featuring routine are dwindling, and are being replaced by a new type of automation - smart machines that learn, and that can gather information, organize it, and produce forecasts from it more effectively and accurately than people.
The employees must adjust to every project
While jobs are disappearing, the demand for employees in the repetitive work of data annotation is continually increasing, because even the most advanced AI technology requires a great deal of manual work and input by people. Machine learning algorithms are based on programming of complex neuron networks fed with millions of annotated pieces of data, and training of these networks is based on these data. Annotating the information is still mostly done manually by people. For example, when Google asks us to state how many stop signs or vehicles appear in a picture in order to make sure that we are not robots, it received annotated images of these objects for training its algorithms.
In order to meet the growing demand for data annotation, an industry has developed in recent years for providing outsourced data annotation services. A 2019 report by research company Cognilytica states that the annotation market totaled $150 million in 2018, and projects growth to $1 billion by 2023. A 2019 report by "The Financial Times" says that hundreds of thousands of workers in the developing countries are employed as low-paid data annotators. They sit for many hours in front of computer screens in crowded offices doing work that they find tiring, exhausting, and unfulfilling.
Point.AI realized that repetitive work requiring concentration and attention to detail is exactly the kind of work that is suitable for a large proportion of the highly functional people on the autistic spectrum. "Right now, we mainly do visual annotation. It's office work, in which the employees sit in front a computer and receive a collection of images and specific requirements of what must be marked in these images. The requirements can be very simple. One especially common example is a collection of aerial photos on which all of the roads or trees must be marked. Another example in fashion is a collection of photos of people in which the types of clothes and people must be marked. This selection and classification work sounds simple, but in other cases, it can be more complicated," Dvir explains.
"We work on a projects model, and our goal is to be a company that pays for itself, pays salaries to its employees, and helps them. We don't want just ordinary customers; we want committed customers with long-term projects," Dvir says. "Since each new project that we get creates a change for the employees, who have to become accustomed to it, the customers must have patience initially in order to obtain the desired results. In time, however, once they get into the routine, the results can be better than expected," she adds.
Dvir says that companies in security sectors that need annotation of sensitive security information cannot use annotation services in foreign countries. Project.AI therefore decided to offer these companies local annotation services. One of Project.AI's five existing customers, whose identity was not disclosed, is in the defense field.
Independence and fair wages
Yalon Shani, 23, one of Point.AI's four employees, has worked at the company for four months. He told "Globes" that he had trouble finding a job since he was demobilized from military service as a reserve noncommissioned officer. He asked for help from Beit Ekstein, a non-profit organization from the Danel group for people with disabilities. Beit Ekstein referred him to an interview with Point.AI, and has continued guiding him since he started working.
"I feel much more fulfilled. I feel that I have meaning in life. I wake up in the morning with something to do. I earn my money by myself, and it feels good to be independent," Shani says. "Everyone in the office has their own quirks, things that you don't expect in an ordinary work environment," he adds. "Some of the projects require less thought and some more, and those are more interesting than the others."
Shani is receives counselling from Beit Ekstein head of placement services Avital Tavor. The placement services are designed for adults on the autistic spectrum. Guidance is provided to 350 adults around Israel, 20% of whom are placed in the technology industry. "Just over six months ago, Eli and Tamar came to us and presented the idea. They wanted to hear from us, people who counsel these adults and are familiar with them and with the difficulties that can arise. They wanted to know how we could help them professionally. It was important for them that the office should be in Sarona, and that the work environment should be that of a startup. They asked us to send them suitable people," Tavor says. "This work environment is not the conventional one for people on the spectrum, because it's located in a noisy area with heavy traffic, which can easily disrupt their focus. But they have succeeded in generating focus there, and in teaching them the job, because the employees don't come from the sector and have no prior training."
She adds that one of the principles is placement at pay rates equal to the average in each job, and Point.AI says that their employees earn an hourly wage in line with the prevailing wage level in the sector.
Academic study programs and an employment apparatus
Tavor says that the big challenge in employing workers on the spectrum is the fact that they are diverse, and share almost no common characteristics. "Every person is different and has his or her own capabilities, but there are frequently gaps in soft expertise, such as the ability to organize a timetable and set priorities, get to work on time, whether to say hello and good morning, how to behave with a manager and colleagues, how to accept criticism and feedback, etc. - everything involving reading social situations," she explains. The necessary adjustments therefore change from one workplace to another, and between individuals.
"More people are diagnosed on the spectrum each year, so we see the number of such people employed in technology, and in general increasing. At the beginning, queries for hiring people on the spectrum in technology were mainly in QA. There were also QA training programs for people with autism, and they received a diploma, but this sector has been automated, and there are fewer jobs," Tavor says.
"Today, there are academic study programs specifically designed for people on the spectrum, and more of them find jobs in technology, with degrees in business administration and computer science. At the same time, we are seeing more and more employers contacting us with jobs for which they want employees," she adds. "One reason is legislation on integration of people with disabilities, but beyond that, people can see the added value in it. We receive a great deal lot of feedback from employers, who say that once they carry out such integration, the work environment becomes 'softer.' The technology environment is fiercely competitive, and employing someone who requires integration makes the environment softer and more inclusive."
Published by Globes, Israel business news - en.globes.co.il - on February 24, 2020
© Copyright of Globes Publisher Itonut (1983) Ltd. 2020