As we continue to spear our heads through Covid we can clearly see that the tech industry did not let the global pandemic get in the way of its progress. In fact, when it comes to AI investments, the word on the street is that what people expected to take years or even a decade to achieve has merely taken a few, albeit strained, months of Covid. The pandemic has been shown to boost the tech’s investment and interest in AI. This year, AI startup funding hit a record high of $17.9B in Q3, according to the latest State of AI report from CB Insights. In addition, in a ManageEngine survey, it was revealed that 80% of their respondents from the US have recently accelerated their AI adoption.
But let’s take a look beyond Covid and assess the year as a whole; where do we stand in the AI industry? What did we gain? What did we learn? Let’s just take a few moments to recap our achievements this year, and then the good stuff on what to expect for 2022.
The State of AI in 2021
Let’s start with the State of AI 2021 Report produced by AI investors, Nathan Benaich, and Ian Hogarth. This whopper of a report (183 slides), highlights what they believe to be AI breakthroughs they’ve witnessed over the past 12 months. I’d like to highlight 3 takeaways I found interesting for 2021.
1. Shifting Focus Bringing Machine Learning to Production:
According to Hazy Research: With the increasing power and availability of machine learning models, gains from model improvements have become marginal. In this context, the machine learning community is growing increasingly aware of the importance of better data practices, and more generally better MLOps, to build reliable machine learning products.
According to Benaich, the shift has moved from just getting a model to work and then worrying about the problems to focusing on a particular problem, and when changes occur, dealing with them. People have acknowledged that machine learning is not static and you have to constantly update it. This acknowledges the notion that datasets are multi-layered and that context is key in delivering quality data.
2. AI Will Transform Drug Discovery & Healthcare:
When biology is run with an AI-first approach, then there are faster simulations of humans’ cellular machinery (proteins and RNA). Some examples include Exscientia, an AI-first drug company. These are the world’s first AI-designed drugs to undergo human testing. AI helped them synthesize 10x fewer compounds enabling them to find a candidate and results in 12 months as opposed to the 54-month industry average.
Another example is Allecyte’s computer vision AI that was able to identify the most powerful drug for each individual cancer patient that would improve their chances of survival. Yet another very promising ongoing research breakthrough was with Shira Barzilay, an MIT researcher who used her own breast cancer survival to devise a technique that seems to predict many breast cancer cases. Barzilay’s research on natural-language processing applies algorithms to textual data. She realized she can put this to a different use: predicting cancer.
3. AI Can Compose Text/Audio/Images On Par To Human-Level Expertise:
AI can now generate content at scale. This opens the possibilities to customizing experiences at speed. Most of you have probably encountered the newly upgraded chatbots. You might find yourself asking, are they human or chatbot? You’re going to have a hard time telling the difference as the bar continues to rise. Not only this, but chatbots are a 24/7 service and since they don’t sleep, people can get support around the clock. In addition, text, audio, and video can be custom-generated content, specific to your liking.
Moving Into 2022
And now onto the good stuff, what is around the corner waiting for us in 2022? Where will AI take us this year? To answer these questions, I’m going to let Dataloop’s founders set the scene.
Note: If you didn’t catch their 2021 predictions, be sure to check that out!
Founders’ Predictions for 2022
Eran Shlomo, Co-founder, and CEO at Dataloop says: “Today’s software users and buyers are used to expecting 100% bug-free software, wherein 100% AI accuracy is not attainable and in many cases even hard to define. In the past years, AI limitations and flaws were viewed as engineering challenges by the AI development communities, while mainstream users had very little interaction with AI-powered applications.
More AI apps will go to production, and this will cause a higher demand for fair & ethical AI. We’ll see a transition from “black box AI” to “gray box AI.” – Eran Shlomo, Co-founder & CEO at Dataloop AI. Click To Tweet
In 2022, many more AI apps will start moving faster towards production and scale, making this new technology’s flaws, limitations, and bugs more broadly discussed by users, businesses, and regulators. This will result in a rising public demand for more explainable, fair, stable, and responsible AI. I see this process unfolding over the next few years, transitioning from “black-box AI” to a “gray-box AI” mindset.” A black box AI app will give unexplainable answers, while gray box AI also supplies key decision points impacting the decision, for example when you see a “why am I seeing this ad?”.
Avi Yashar, Co-founder and Chief Product Officer at Dataloop says: “2022 is going to be the year that traditional businesses will start to embrace AI solutions to be more competitive and efficient. Companies are already showing this level of understanding and to reiterate the ManageEngine survey, 80% of American companies have accelerated their AI adoption. Furthermore, I believe that the next wave of AI will be human augmented AI. Humans and machines will work together to help businesses be more effective. Human in the loop or human augmented AI, involving humans not only refers to the initial training phase of machine learning models but also to the real-world application of AI models. This allows humans to stay in the loop and augment with AI and essentially fine-tune, validate and deliver the most accurate AI model to the end-user.”The next wave of AI will be human augmented. Humans and machines will work together to help businesses be more effective.” – Avi Yashar, Co-Founder & Chief Product Officer Dataloop AI. Click To Tweet
Nir Buschi, Co-founder and Chief Business Officer at Dataloop says: “In 2022, human intelligence will make its way into the AI data development technology stack as yet another fully integrated component.”
We will see a shift from the slap-on approach in the industry of contracting or outsourcing external data labelers as a source for human intelligence to (what has already started) and will continue to pick up to a full-stack “human-in-the-platform-loop” methodology. A best-of-breed end-to-end MLOps infrastructure and toolset together with human-centric data operations create a unifying data lifecycle experience, all under one roof.”
At Dataloop we’re looking forward to 2022. We have hope, optimism, and foresight that this year will be a note for the books in AI. You’ve seen the surveys, the reports, and heard our predictions. Now, all we can do is sit back and watch the year unfold and we’ll pick it up next year to reassess… We’d love to hear your AI predictions in 2022, and also what you think of our predictions: click here to share your insights. Wishing everyone a healthy and prosperous year!