Machine learning is rapidly transforming various industries, from recommending your next favorite movie on Netflix to powering complex medical diagnoses. It’s becoming so integral to artificial intelligence (AI) that the terms are often used interchangeably. But what exactly is machine learning, and how is it relevant to everyday technologies, like the computers in our cars? While the question “Do Used Computers Need To Be Programmed In Cars?” might seem straightforward, it opens up a broader discussion about the sophistication of modern automotive systems and the role of machine learning within them.
In essence, machine learning empowers computers to learn from data without explicit programming. MIT Sloan professor and founding director of the MIT Center for Collective Intelligence, highlights this shift, noting that machine learning has become “arguably the most important way” AI is developed today. This explains why AI and machine learning are often seen as synonymous – most recent advancements in AI are driven by machine learning.
The prevalence of machine learning is undeniable. A 2020 Deloitte survey revealed that 67% of companies already utilize machine learning, with a staggering 97% planning to implement or expand its use in the near future. From optimizing manufacturing processes to personalizing retail experiences, machine learning is reshaping industries across the board. As MIT computer science professor and director of the MIT Center for Deployable Machine Learning, Aleksander Madry, emphasizes, “Machine learning is changing, or will change, every industry, and leaders need to understand the basic principles, the potential, and the limitations.”
Understanding machine learning isn’t just for tech experts. It’s crucial for anyone in business to grasp its fundamental principles, capabilities, and constraints. This includes recognizing the societal and ethical implications of this powerful technology. As Dr. Joan LaRovere, a pediatric cardiac intensive care physician and co-founder of The Virtue Foundation, aptly puts it, “AI has so much potential to do good, and we need to really keep that in our lenses as we’re thinking about this. How do we use this to do good and better the world?”
Decoding Machine Learning: How Computers Learn
Machine learning is a specialized field within artificial intelligence. AI, broadly speaking, is the ability of a machine to mimic intelligent human behavior, tackling complex tasks in a human-like problem-solving manner. Boris Katz, a principal research scientist and head of the InfoLab Group at CSAIL, explains that AI aims to create computer models exhibiting “intelligent behaviors” such as visual scene recognition, natural language understanding, and physical world interaction.
Machine learning is a key approach to achieving AI. As defined by AI pioneer Arthur Samuel in the 1950s, it’s “the field of study that gives computers the ability to learn without explicitly being programmed.” This definition remains accurate today. Imagine traditional programming, or “software 1.0,” as baking with a strict recipe. You meticulously follow instructions with precise measurements and timings. Traditional programming similarly demands detailed instructions for a computer to execute.
However, creating explicit programs becomes challenging or even impossible for complex tasks like teaching a computer to recognize faces. While humans effortlessly identify faces, translating this ability into computer code is incredibly difficult. Machine learning offers an alternative: it allows computers to learn through experience, programming themselves from data.
This learning process begins with data – it could be numbers, images, or text, such as financial transactions, photos of individuals, or sensor data. This data is meticulously collected and prepared as “training data,” the information the machine learning model will learn from. The more data available, the better the program’s learning potential.
Programmers then select an appropriate machine learning model, feed it the training data, and allow the model to train itself, identifying patterns and making predictions. Human programmers can refine the model over time, adjusting parameters to enhance accuracy. Janelle Shane’s website, AI Weirdness, offers a humorous perspective on machine learning algorithms’ learning curves and occasional missteps, like the time an algorithm attempted to generate recipes and produced “Chocolate Chicken Chicken Cake.”
To evaluate the model’s performance, some data is set aside as “evaluation data.” This data tests the model’s accuracy when exposed to new, unseen information. The outcome is a trained model ready for future use with different datasets.
Successful machine learning algorithms can fulfill various functions, as outlined in a research brief on AI and the future of work co-authored by MIT Professor Daniela Rus, Robert Laubacher, and Thomas Malone. These functions can be:
- Descriptive: Explaining past events based on data.
- Predictive: Forecasting future outcomes using data patterns.
- Prescriptive: Recommending actions based on data analysis.
Machine learning is further categorized into three main types:
-
Supervised Learning: Models are trained using labeled datasets, enabling them to learn and improve accuracy over time. For example, an algorithm trained with labeled images of dogs and other objects learns to identify dogs independently. This is the most prevalent type of machine learning today.
-
Unsupervised Learning: Programs analyze unlabeled data to discover hidden patterns. This can reveal trends or groupings that might not be immediately apparent to humans. For instance, unsupervised learning could analyze sales data to identify distinct customer segments.
-
Reinforcement Learning: Machines learn through trial and error, guided by a reward system to encourage optimal actions. This approach is used to train game-playing AI or autonomous vehicles, rewarding correct decisions to facilitate learning over time.
Infographic illustrating the process of machine learning, highlighting data input, model training, and insight generation.
Malone’s Work of the Future brief emphasizes that machine learning thrives on vast datasets – thousands or millions of examples such as customer conversation recordings, sensor logs, or transaction records. Google Translate’s success, for example, stems from its training on the immense amount of multilingual information available online.
Machine learning can unlock insights and automate decisions in scenarios beyond human capabilities, as Madry points out. “It may not only be more efficient and less costly to have an algorithm do this, but sometimes humans just literally are not able to do it,” he states.
Google Search exemplifies this. While humans can search, the scale and speed at which Google’s machine learning models deliver results are unmatched. Malone clarifies, “That’s not an example of computers putting people out of work. It’s an example of computers doing things that would not have been remotely economically feasible if they had to be done by humans.”
Machine learning also intersects with other AI subfields:
-
Natural Language Processing (NLP): NLP empowers machines to understand and process human language, both spoken and written, unlike traditional computer programming that relies on structured data. This enables technologies like chatbots and virtual assistants like Siri and Alexa to understand and respond to human language.
-
Neural Networks: These are a common type of machine learning algorithm inspired by the human brain. Artificial neural networks consist of interconnected processing nodes organized in layers. Data flows through these nodes, with each node processing inputs and generating outputs for subsequent nodes. In image recognition, for example, different nodes analyze features to determine if an image contains a specific object.
-
Deep Learning: Deep learning networks are neural networks with multiple layers. This layered structure allows for processing massive datasets and assigning “weights” to connections within the network. In image recognition, some layers might identify basic features like edges, while deeper layers recognize complex patterns like faces. Deep learning, mimicking the human brain, powers many advanced AI applications, including self-driving cars and medical diagnostics. Malone notes, “The more layers you have, the more potential you have for doing complex things well.” However, deep learning’s high computational demands raise concerns about its economic and environmental sustainability.
Machine Learning in Action: Business Applications
For some companies, machine learning is the core of their business model, like Netflix’s recommendation system or Google’s search engine algorithms. Other businesses are deeply integrating machine learning, even if it’s not their primary offering. Many more are still exploring how to leverage machine learning effectively. Shulman observes, “In my opinion, one of the hardest problems in machine learning is figuring out what problems I can solve with machine learning. There’s still a gap in the understanding.”
Researchers at the MIT Initiative on the Digital Economy developed a 21-question rubric in a 2018 paper to assess task suitability for machine learning. Their findings indicate that while no occupation will be untouched by machine learning, complete automation of entire occupations is unlikely. The key to successful machine learning implementation lies in breaking down jobs into individual tasks, some suitable for machine learning and others requiring human expertise.
Companies are currently applying machine learning in diverse ways:
-
Recommendation Algorithms: Powering suggestions on platforms like Netflix, YouTube, Facebook, and e-commerce sites, these algorithms learn user preferences to deliver personalized content and product recommendations. Madry explains, “[The algorithms] are trying to learn our preferences. They want to learn, like on Twitter, what tweets we want them to show us, on Facebook, what ads to display, what posts or liked content to share with us.”
-
Image Analysis and Object Detection: Machine learning analyzes images for information like identifying and differentiating individuals. While facial recognition technology raises ethical concerns, business applications are emerging. For instance, hedge funds use machine learning to analyze parking lot occupancy from satellite images, providing insights into company performance.
-
Fraud Detection: By analyzing spending patterns and typical behavior, machine learning algorithms can identify potentially fraudulent credit card transactions, suspicious login attempts, and spam emails.
-
Automatic Helplines and Chatbots: Companies are deploying chatbots for customer service, using natural language processing and machine learning to learn from past conversations and provide automated responses.
-
Self-Driving Cars: A significant portion of self-driving car technology relies on machine learning, particularly deep learning, to navigate roads and make driving decisions.
-
Medical Imaging and Diagnostics: Machine learning programs are trained to analyze medical images like mammograms to detect subtle markers of disease, such as predicting cancer risk.
[
Leading the AI-Driven Organization
In person at MIT Sloan
Register Now
Machine Learning: Promises and Challenges
While machine learning drives innovation and efficiency, business leaders must be aware of its limitations and potential pitfalls.
Explainability:
A key concern is “explainability” – understanding how machine learning models arrive at decisions. Madry stresses, “Understanding why a model does what it does is actually a very difficult question, and you always have to ask yourself that. You should never treat this as a black box… try to get a feeling of what are the rules of thumb that it came up with? And then validate them.”
This is crucial because machine learning systems can be vulnerable to manipulation or fail unexpectedly, even in tasks easily performed by humans. For example, subtle alterations to image metadata can mislead AI, causing a machine to misidentify a dog as an ostrich.
Madry highlights an example where a machine learning algorithm analyzing X-rays seemed to outperform doctors in diagnosing tuberculosis. However, further investigation revealed the algorithm was correlating diagnoses with the age of the X-ray machine, not the image itself. Older machines, more common in developing countries with higher tuberculosis rates, skewed the results. The algorithm achieved the task, but not in a clinically meaningful way.
The importance of model explainability and accuracy varies depending on the application. While 95% accuracy might be acceptable for movie recommendations, it’s insufficient for critical applications like self-driving vehicles or machinery fault detection.
Bias and Unintended Outcomes:
Machine learning models are trained on data, and if this data reflects existing biases, the algorithms will learn and perpetuate these biases. For example, chatbots trained on Twitter conversations might adopt offensive language.
In some cases, machine learning can exacerbate societal issues. Facebook’s use of machine learning to personalize content, while aiming for engagement, has inadvertently contributed to the spread of extreme content, polarization, and conspiracy theories by prioritizing sensational and often inaccurate information.
Addressing bias in machine learning requires careful vetting of training data and organizational commitment to ethical AI practices. This includes embracing human-centered AI, which emphasizes diverse perspectives in AI system design. Initiatives like the Algorithmic Justice League and The Moral Machine project are actively working to promote fairness and ethical considerations in AI.
Leveraging Machine Learning Effectively
Shulman points out that executives often struggle to identify where machine learning can genuinely add value. What’s a gimmick for one company might be core to another. Businesses should focus on practical applications aligned with their specific needs, rather than blindly following trends.
The successful machine learning strategies of companies like Amazon may not directly translate to a car manufacturer. While voice assistants are central to Amazon’s ecosystem, a car company might find more impactful applications in optimizing factory processes using machine learning. Considering the automotive industry, the question of “do used computers need to be programmed in cars?” becomes relevant in the context of retrofitting older vehicles or utilizing existing hardware for new functionalities. While modern cars increasingly rely on powerful embedded systems and sophisticated software, understanding the programming capabilities and limitations of “used computers” – perhaps referring to older or less powerful ECUs – is crucial for cost-effective upgrades or repairs. The complexity of modern vehicle systems, especially with the integration of machine learning for advanced driver-assistance systems (ADAS) and autonomous driving features, necessitates careful consideration of hardware and software compatibility.
Shulman advises against viewing machine learning as a solution searching for a problem. Instead of starting with technology, businesses should identify business problems or customer needs that machine learning can effectively address.
LaRovere emphasizes that while a basic understanding of machine learning is important, successful implementation requires collaboration across different areas of expertise. “I’m not a data scientist… but I understand it well enough to be able to work with those teams to get the answers we need and have the impact we need,” she states. “You really have to work in a team.”
Learn More:
Explore resources to deepen your understanding of machine learning:
- Sign-up for a Machine Learning in Business Course.
- Watch an Introduction to Machine Learning through MIT OpenCourseWare.
- Read about how an AI pioneer thinks companies can use machine learning to transform.
- Watch a discussion with two AI experts about machine learning strides and limitations.
- Take a look at the seven steps of machine learning.
Read next: 7 lessons for successful machine learning projects
For more info Sara Brown Senior News Editor and Writer [[email protected]](mailto:[email protected] "email at [email protected]")[
