AI Is Eating Software deepkapha.ai
By Martijn van Attekum, Jie Mei and Tarry Singh
Introduction
Marc Andreessen famously said that “Software is eating the world” and everyone gushed into the room. This was as much a writing on the wall for many traditional enterprises as it was wonderful news for the software industry.
Still no one actually understood what he meant.
To make his point he stated this example:
"Today, the world’s largest bookseller, Amazon, is a software company — its core capability is its amazing software engine for selling virtually everything online, no retail stores necessary. On top of that, while Borders was thrashing in the throes of impending bankruptcy, Amazon rearranged its web site to promote its Kindle digital books over physical books for the first time. Now even the books themselves are software."
This was 2011.
Marc Andreessen TechCrunch
Interestingly, Andreessen also said the following:
"I, along with others, have been arguing the other side of the case...We believe that many of the prominent new Internet companies are building real, high-growth, high-margin, highly defensible businesses."
(Read the full blog article at his a2z VC fund)
Little did Andreessen envision that the same software industry could be at risk of being eaten.
Fast forward to 2019 and the very same software industry is nervous. Very very nervous!
And the reason is AI.
Especially for those who haven’t bulked up their AI warchest.
Acceleration Wave (2009 - 2019) - When Software Started Eating the World
Andreessen was right.
The companies that embraced software in 2011 are the current market leaders in their respective fields, and the top 5 market capitalization companies worldwide in the second quarter of 2019 are all offering some type of software solutions (ycharts.com).
Concurrently, the period since 2011 has shown an unprecedented growth in the developments in AI. Although several key ideas about AI have been around for long, a number of processes have accelerated their potential use.
First, computing power, in particular for specialized AI chipsets, has vastly increased.
Second, the amount of training data for AI algorithms is exploding with the advent of data lakes and a fully connected internet-of-things world, expanding AI domains and decreasing the costs to train algorithms.
Third, a large number of technological bottlenecks (such as vanishing gradients) have been solved over the last few years, massively increasing accuracy and applicability of existing algorithms.
Lastly, the decrease in costs for cloud storage and computing plus the facilitation of distributed collaborative working, made combining highly specialized knowledge easier than ever before.
The extent in which Andreessen’s cherished software companies are weaving AI into their products is however often limited. Instead, a new slew of start-ups now incorporates an infrastructure based around the above mentioned AI-facilitating processes from their very foundation.
HyperAcceleration Wave (2019 - 2030) - AI Has Started Eating Software
Driven by an increase in efficiency, these new companies use AI to automate and optimize the very core processes of their business. As an example, no less than 148 start-ups are aiming to automate the very costly process of drug development in the pharmaceutical industry according to a recent update on BenchSci.
Likewise, AI start-ups in the transportation sector create value by optimizing shipments, thus vastly reducing the amount of empty or idle transports.
Also, the process of software development itself is affected. AI-powered automatic code completion and generation tools such as TabNine, TypeSQL and BAYOU, are being created and made ready to use.
Let’s quickly look at a few example applications of this hyperacceleration wave:
Automating the coding process
DeepTabNine Tabnine
It is trained on around 2 million files from code repository GitHub. During training, its goal is to predict each token given the tokens that come before it. To achieve this goal, it learns complex behaviors, such as type inference in dynamically typed languages.
Once Deep TabNine developers realized the parallel between code and natural language processing, they implemented the existing GPT-2 tool which uses the Transformer network architecture.
The inventor of this tool is Jacob Jackson, an undergraduate student and ex-OpenAI intern who quickly realized this idea and created a software tool for it.
Getting answers to any question about your medical data
As AI will create the query to get the answer for you!
Here, a group of medical researchers created a tool that you can ask literally any questions on medical data and the AI generates a customized SQL query that is then used to retrieve the relevant data from the database.
Speech Text to Generating Database Query automatically Question to SQL Generation
It's called Question-to-SQL generation.
They used RNN (a form of deep learning, an AI on steroids for text analytics) with Attention and Point-Generator Network. For those more inclined to exploring the technical part of this feel free to read their research here and software code here.
So is it time the armies of database administrators (DBAs) to go home?
Creating a beautiful website based on your sketch
While AI translates your sketch into code!
Want to build your website quickly? All you need to do is sketch it and this platform will use AI to create software code like html, css and js code ready in vue.js instantly.
Sketch to create a website with AI Zecoda
Easy, huh?
Just input your sketch and voila! your website pops out at the other end!
Find out more about this platform here.
These are just a few examples of how AI is increasingly encroaching all parts of software development and eliminating mundane tasks of coding and programming rapidly!
This is due to the motivation to automate the process of numerical analysis, data collection and eventually, processing and relevant code production.
Researchers have higher-than-ever awareness and knowledge to infiltrate each and every problem at all levels with AI-powered software, from day-to-day anecdotes such as: Which kind of cookies shall we recommend to a customer given their shopping preferences?
To large-scale, manufacturer’s dilemma, for example:
How do we automate the production line in an individualized yet systematic manner?
And finally, to the processing of building smarter, easier-to-use software that may even write code for you.
Apart from assisted decision making, diagnostic and prediction, work of AI researchers and influencers have led to a hyperacceleration wave: Software powered by AI does not only achieve performances comparable to the human level, but creates something that would challenge an average person’s imagination and perception of their own abilities.
A person can no longer tell apart the fake celebrity faces generated by generative neural networks from the real ones, or need not remember the name of every function they will use when writing a script.
Imaginably, the wide application domains and near-human performance of AI-powered software will cause a paradigm shift in the way people deal with their daily personal and professional problems.
Although some of us are pessimistic about, or in some extreme cases, consciously avoiding a world with overwhelming AI-powered software, there is not so much room for an escape. Amazon, Google, and even your favorite neighborhood florist, are actively (and sometimes secretly) using AI to generate revenue. Face it, or be left behind.
What would you do if you were BMW today?
"At this point, no one can reliably predict how quickly electromobility will progress, or which drive train will prevail... There is no customer requests for self driving BEVs. (electric vehicles)"
A classic trap most big enterprises with established business fall for is getting micro-focused on existing business segments while losing sight on the slowly eroding economic and business climate.
Tesla's story as an electric car is known to all but many may not know that it is the self-driving feature and the heavy use of AI in both software and hardware where the secret sauce lies.
They have already driven 10 billion electric miles and the cars are collecting all the more data to disrupt not just the automotive markets but its adjacent markets in manufacturing, servicing, sales and in general mobility.
Tesla's AI is eating all other automotive industry's business.
A few weeks later after his annual address, the BMW chief had resigned.
CEO's and executives who however do wish to proactively adopt AI should do the following 5 things
Concluding thoughts
1) Have your AIPlaybook Ready
Last year I did a keynote panel together with a few industry peers and I was asked if AI could eat software and I said "Yes".
Take a listen.
Any company that is not in possession of its AI Playbook, that is not armed with data, algorithms and machine learning models, is certainly going to find itself in serious quandary.
An example of an AI playbook is to assess your firm's maturity thoroughly and plan for ROI driven projects.
AI Playbook deepkapha.ai
2) Upskill and/or hire a (good) data science team
Upskilling your staff to be able to drive your AI transformation is the key to success for any organization aspiring to become an AI company.
We've advised several large-scale data-intensive projects and here are a couple of key arguments that executives should take to heart.
- In a couple of years embracing AI is not a matter of trend riding, but survival;
- To survive an era in which AI is dominating both market and software, CEOs and executives need to level up their mindset for successful adoption and application of AI within their enterprise, for which they either have to upskill or find a good data science team;
- Know your game: A good team helps you understand how AI will make your company survive;
- Examples are abundant in the industry and it is key for companies to pay attention to latest trends and launch several smaller projects to extract out the key projects that can be industrialized at scale.
3) Develop Algorithms & Execute Your Data-Play From Day 1
Upgrading your technical infrastructure that can develop the latest AI algorithms, process large quantities of heterogenous datasets, build and train both industry benchmarked and novel AI models is an important first step.
Once that is established it is very critical to develop meaningful dialog channels to envision and dream project ideas that are pain killers and dive directly into solving those problems with data.
Finally, executing from Day 1 on the "good-enough" data models and algorithms is where a true AI company can define its momentum and gain sizeable lead from its nearest competition.
4) Implement a distributed knowledge structure
As access to the right data is a key to valuable AI solutions, ensuring access to data generated or acquired within the company and outside will be of crucial importance. Following this realization, pharmaceutical companies are starting to create central repositories of the data gathered in their clinical trials. Consequently, their data science teams will have access to a structured knowledge database they can use to train AI algorithms.
A second way to ensure the distribution of knowledge, is to set up a distributed collaboration structure. With the advent of software mimicking group processes from setting schedules, having meetings, or doing a brainstorming session, integration of knowledge and expertise should no longer be limited by geographical location.
5) Tap into AI start-ups with relevant knowledge
Andreessen’s example of Disney buying Pixar in order to stay relevant has paid off for Disney, which sold for over 8 billion dollar in movie tickets this year, making Disney the second biggest media company (Forbes).
Yet the latest developments suggest AI could also optimize movie-making processes. Moreover, as Disney is creating a consumer platform with Disney+, AI might form the necessary basis to ensure optimal usage of the data generated by this platform. When not wanting to build data science teams from scratch, collaborating with or taking over relevant start-ups might again be necessary for companies such as Disney to stay competitive.
So yes, AI has started eating software.
What are you going to do?
___________________________________________________________
About contributing authors
Martijn v Attekum MD (Oncology) and PhD
Dr. Martijn Van Attekum (MD, PhD) works as a data scientist in biomedicine at the University of Cologne. He is an experienced project manager and writer, and is skilled in genomics, oncology, and machine learning. As Visiting AI Researcher at deepkapha.ai he participates in ground-breaking deep learning projects on medical image analysis. In his free time, he is very much attracted to everything the mountains have to offer, such as climbing, hiking, and mountain biking.
Jie Mei PhD Computational Neuroscience
Dr. Jie Mei is a computational neuroscience researcher who has completed her studies at the Ecole normale supérieure and Charité Universitätsmedizin Berlin. She is currently based in Edmonton, Canada and is responsible for the growth of AI research department within deepkapha.ai and its companies. Her research interests include computational neuroscience, neurorobotics, machine learning and data analytics in healthcare and medicine. She is also an active startup advisor.
/w=3840,quality=90,fit=scale-down)
