Human-in-loop to make the AI applications possible

Reading Time: 3 minutes |

September 20, 2022




Human-in-loop to make the AI applications possible

Artificial intelligence is gaining new grounds. The use of AI is increasing in various traditional fields like monitoring on the factory floor, automating customer service communication, analyzing vast amounts of data generated for business intelligence, marketing, operations covering almost all aspects of a business. It has also given rise to new areas like autonomous vehicles, computer vision-based surveillance and even AI-generated music and paintings!

This is truly the golden age of artificial intelligence with great amounts of investments, research and adoption in real-world use cases. The boom started with deep learning and specifically neural networks becoming easy to train with the advancements in processing speeds of hardware as well as a large amount of data storage and transfer now possible with the maturing cloud ecosystem. But, most of the AI in the application today is still based on supervised learning. This includes both natural language processing and computer vision applications. This means that for the machine to get better at classification, identification or any other task that the machine learning system is built for, the system needs to train on a large amount of labelled data.

For example, if the system is built for detecting pedestrians or other obstacles on the road in case of autonomous driving, the system needs to be trained on a vast amount of image data which already has the pedestrians marked as pedestrians or trash cans marked as trash cans usually by humans. This can be quite cumbersome and time-consuming as the system needs a substantially large amount of labelled data to become acceptably accurate.

AI researchers have been working on other forms of learning like reinforcement learning as well as completely unsupervised learning methods. Reinforcement learning has achieved decent progress, but is suitable for only certain kind of applications. Unsupervised learning is still in the research phase and still requires development to be considered for applied AI applications. Therefore, the current AI applications in the commercial field usually depend on supervised learning for almost all practical purposes. And with the rising number of applications, there is a huge demand for labelled data-sets for a wide range of applications like automated responses in both writing and voice, autonomous driving, emotion recognition, etc.

All this has led to a boom in new-age startups providing training data or labelled data as a service. What is means is that humans are making artificial intelligence smarter in the background!! A somewhat interesting way in which AI is creating jobs for a completely new need! What is also interesting to note that quite a few of these companies are ‘social impact’ employers as a lot of this work can be done remotely. Companies are hiring in parts of Africa as well as women in disadvantaged communities to do these jobs remotely without having to leave their houses. Many of them are using AI themselves to streamline the processes and maintain the quality of annotation of data. These AI-training-as-a-service startups have gathered a lot of attention as well as a lot of funding in recent times.

Scale (2016, San Francisco) is one such startup closed a $100 million Series C round of financing led by Founders Fund with participation from Accel, Coatue Management, Index Ventures, Spark Capital, Thrive Capital, Instagram founders Kevin Systrom and Mike Krieger and Quora CEO Adam D’Angelo. The startup founded in 2016 has already raised USD 22.5 million earlier and has managed to gather an enviable list of customers in the autonomous driving space with almost all the major companies like Waymo, Cruise, Zoox, Lyft, Toyota, Uber and others as well as other marquee names like P&G, Mapbox, OpenAI, Pinterest, Airbnb, etc. The company uses an API model which allows companies to directly submit their AI training tasks and the company uses a combination of human resources along with intelligent tools to annotate the data. Scale AI is now considered to be the latest unicorn in the silicon valley.

Only a few months back, Appen acquired Figure Eight (earlier known as CrowdFlower) for USD 300 million, making it one of the largest data annotation companies with over 1 million remote workers. Appen had earlier acquired a similar company, Leapforce in 2017. There are other startups in the AI-training-as-a-service space which also have received significant funding and have a decent customer base. These include companies like CloudFactory, iMerit, Mighty AI, Samasource and others.

With the rising application of AI in industries, this is one space to watch out for in the future. The development in unsupervised learning should also be closely watched as the current dependence on complete supervised learning may not be very fruitful for much smarter applications in AI. The goal of artificial general intelligence is still far away and needs a paradigm shift in the present research. Till then, the human will continue to power artificial intelligence as well!

To deep dive and stay continuously updated about the most recent global innovations in Artificial Intelligence and learn more about applications in your industry, test drive WhatNext now!

How useful was this post?

Click on a star to rate it!

Average rating 0 / 5. Vote count: 0

No votes so far! Be the first to rate this post.

Leave a Comment

Your email address will not be published. Required fields are marked *

Related Insights

Food Supply Chain - WhatNext

Food Supply Chain and Internet of Things

Driver Monitoring using AI -WhatNext

Driver Monitoring using Artificial Intelligence

Quantum Computing - WhatNext

Quantum Computing in Car Manufacturing

Sustainable Agriculture - WhatNext

Sustainable Agriculture using Synthetic Biology

Potential of Living Medicines - WhatNext

Potential of Living Medicines