Intro
Marc Andreessen famously said, “Everyone is pouring into the room. For many traditional businesses, this is also a good writing, because this is good news for the software industry.
Still no one really understands what he means.
To illustrate his point, he said this example:
"Today, the world’s largest bookseller, Amazon, is a software company – its core function is its amazing software engine, which can sell almost everything online, without the necessary retail stores. Most importantly, when Borders is about to When struggling in the throes of bankruptcy, Amazon rearranged its website to promote its Kindle digital books through physical books for the first time. Now, even these books themselves are software.
This is the 2011 year.
Interestingly, Anderson also said the following:
"I and others have been arguing about the other side of the case... We believe that many famous new Internet companies are building real, high-growth, high-profit, highly defensive businesses."
(read hisa2z VC FundComplete blog post)
Anderson rarely thinks that the same software industry may be eaten.
Fast forward to 2019, the same software industry is also very nervous. very nervous!
The reason is artificial intelligence.
Especially those who did not increase the AI war.
The acceleration wave (2009-2019)-when software starts to eat the world
Anderson is right.
2011's software-based companies are the current market leaders in their respective fields, and 2019's top five global market capitalization companies offer some type of software solution in the second quarter of the year (Ycharts.com ).
At the same time, the development of artificial intelligence has shown unprecedented growth since 2011. While several key ideas about artificial intelligence have existed for a long time, many processes have accelerated their potential use.
First, computing power, especially for dedicated AI chipsets, has increased dramatically.
Secondly, with the emergence of data lakes and fully connected IoT worlds, the amount of training data for artificial intelligence algorithms is increasing, expanding the field of artificial intelligence and reducing the cost of training algorithms.
Third, a number of technical bottlenecks (such as vanishing gradients) have been resolved in the past few years, greatly improving the accuracy and applicability of existing algorithms.
Finally, the cost reduction of cloud storage and computing and the promotion of distributed collaboration make it easier to combine highly specialized knowledge than ever before.
However, the extent to which Anderson's cherished software companies weave AI into their products is often limited. Instead, a large number of start-ups are now integrating infrastructure based on the artificial intelligence-promoting processes described above.
HyperAcceleration Wave (2019 – 2030)-AI has begun to eat software
Driven by increased efficiency, these new companies use AI to automate and optimize the core processes of their business. For example, according toBenchSciLatest update,at leastThe goal of 148 startups is to automate the very expensive drug development process in the pharmaceutical industry.
Similarly, artificial intelligence start-ups in the transportation sector create value by optimizing freight, thereby greatly reducing the amount of idle or idle transportation.
In addition, the process of software development itself is also affected. Artificial intelligence-driven automated code completion and generation tools such asTabNine .TypeSQL和BAYOU, is being created and ready to use.
Let's take a quick look at some sample applications for this super-acceleration wave:
Automated coding process
By letTabNine Use AI to automate your code!
It is trained by approximately 200 files from the code base GitHub. During training, the goal is to predict each token given the previous token. To achieve this goal, it learns complex behaviors such as type inference in dynamically typed languages.
Once Deep TabNine developers realize the parallel between code and natural language processing, they are usedTransformExisting GPT-2 tools for network architecture.
The inventor of this tool isJacob Jackson , an undergraduate and formerOpenAIThe intern, he quickly realized the idea and created a software tool for it.
Get answers to any questions about your medical data
Because AI will create a query to get an answer!
Here, a group of medical researchers created a tool that allows you to query any questions about medical data verbatim, and the AI generates a custom SQL query that is then used to retrieve relevant data from the database.
It is called Q&A SQL generation.
They use the attention and point generator networkRNN(A form of deep learning, AI for steroids for text analysis). For those who are more inclined to explore the technical part, you canHereRead their research and software代码.
So when does the database administrator (DBA) army go home?
Create a beautiful website based on the sketch
AI converts your sketches into code!
Want to quickly build your website? All you need to do is sketch it, this platform will use AI to create software code immediately in vue.js, such as html, css and js code.
Ok, right?
The input side only needs your sketch adverbial clause:瞧! Your website pops up on the other end!
Here atLearn more about the personal library on this platform.
These are just a few examples of how AI is increasingly encroaching on all parts of software development and quickly eliminating the tedious task of coding and programming!
This is due to the motivation for automated numerical analysis, data collection, and final processing and related code generation processes.
Researchers have more knowledge and knowledge than ever before, and can permeate every problem at every level through artificial intelligence software. It can be seen from everyday anecdotes: according to their shopping preferences, we should recommend to customers. Which kind of biscuit?
For the dilemma of large-scale manufacturers, for example:
How do we automate our production lines in a personalized but systematic way?
Finally, deal with building smarter, easier-to-use software and even writing code for you.
In addition to assisted decision making, diagnosis, and prediction, the work of artificial intelligence researchers and influencers has led to ultra-fast waves: software driven by artificial intelligence not only achieves performance comparable to human level, but also creates some imaginations that can challenge ordinary people. Force and perception of things. Their own abilities.
One can no longer distinguish the fake celebrity faces generated by the neural network, or the names of each function they will use when writing the script.
It is conceivable that the wide range of applications and near-humanized performance of artificial intelligence software will lead to a shift in the way people handle everyday personal and professional problems.
While some of us are pessimistic about consciously avoiding the world of overwhelming artificial intelligence software, or in some extreme cases, there is not much room to escape. Amazon, Google, and even your favorite neighbor's flower shop are actively (and sometimes secretly) using artificial intelligence to generate revenue. Face it, or be left behind.
If you are a BMW today, what would you do?
"At this point, no one can reliably predict how quickly electric vehicles will progress, or which drivetrain will prevail... No customer is required to drive BEVs on their own. (Electric vehicles)"
A classic trap Most large companies with mature businesses are focusing on existing business areas while ignoring the slow erosion of the economic and business environment.
The story of Tesla as an electric car is well known, but many people may not know that this is autonomous driving and the use of artificial intelligence in software and hardware.
They have already boosted 100's miles of electricity miles, and the car is collecting all the more data, not only breaking the automotive market, but also breaking manufacturing, service, sales and general mobility in adjacent markets.
Tesla's AI is consuming the business of all other automotive industries.
A few weeks later, after his annual speech, BMWThe CEO resigned.
CEOs and executives who want to take the initiative to adopt AI should do the following 5 events.
Final thought
1) ready for AIPlaybook
Last year, I worked with some of my peers to make a keynote speech group. Someone asked me if AI can eat software. I said "yes".
listen
Any company that does not have an artificial intelligence manual, without data, algorithms and machine learning models, will certainly be in serious trouble.
An example of an AI manual is a thorough evaluation of a company's maturity and planning of ROI-driven projects.
2) Upskill and / or hire (good) data science team
Enhancing your employees' ability to drive your AI transformation is key to any organization that is interested in becoming an AI company.
We've provided recommendations for several large data-intensive projects. Here are a few key arguments that executives should consider.
- Embracing artificial intelligence in a few years is not a trend, but survival;
- In order to survive the era of artificial intelligence dominated by markets and software, CEOs and executives need to improve their thinking styles in order to successfully adopt and apply artificial intelligence within the enterprise. They must either upgrade their skills or find a good data. Science team
- Know your game: A great team can help you understand how AI can make your company survive;
- There are plenty of examples in the industry, and it is critical that companies focus on the latest trends and launch several smaller projects to extract key projects that can be industrialized on a large scale.
3) Develop algorithms and perform data playback from 1 days
Upgrading your technology infrastructure to develop the latest AI algorithms, processing large heterogeneous data sets, building and training industry benchmarks and novel AI models is an important first step.
Once established, it is critical to develop meaningful dialogue channels to conceive and dream of project ideas as painful killers and directly sneak into data issues.
Finally, starting with the 1 days to implement "good enough" data models and algorithms, real AI companies can define their dynamics and gain considerable lead from recent competition.
4) implements distributed knowledge structure
Since accessing the right data is key to a valuable AI solution, it's important to ensure that data is generated or acquired inside and outside the company. After achieving this goal, pharmaceutical companies began creating central repositories for the data they collected in their clinical trials. As a result, their data science team will have access to a structured knowledge database they can use to train AI algorithms.
The second way to ensure knowledge distribution is to build a distributed collaboration structure. As software mimics team process settings, from setting up schedules, holding meetings or brainstorming sessions, the integration of knowledge and expertise is no longer limited by geographic location.
5) Use relevant knowledge to mine AI startups
Anderson's example of buying Pixar for Disney to maintain relevance has paid off Disney's return. Disney's movie fare this year exceeds 80 billion, making Disney the second largest media company (Forbes).
However, the latest developments show that AI can also optimize the film production process. In addition, because Disney is using Disney+ to create consumer platforms, artificial intelligence may be the necessary foundation to ensure optimal use of data generated by this platform. If you don't want to build a data science team from scratch, companies like Disney may need to work with related startups or take over startups to stay competitive.
Yes, AI has started eating software.
what are you going to do?
This article is reproduced from Forbs,Original address
Comments