top of page

Jacob's A.I. Journal: Vol. V

Jacob's A.I. Journal is a collection of articles on A.I. topics from's Principal ML/AI Architect, Jacob Haning


A.I. in Business

How the energy sector is using Artificial intelligence to digitize their industry

Chevron sought to digitize their document processing, from information extraction to contract optimization. These use cases as outlined in the article show that the growing trend in AI adoption across industries is in digitizing operations. When rules-based automation is insufficient, augmented decision-making software can be used to help solve real scalability and efficiency issues within most sectors. Outside of big-tech I have yet to see AI truly transform industries, as most believe it eventually will. Until then, finding true value from the billions being invested in AI and Machine Learning departments is crucial to keeping momentum and garnering trust from executives. It has been suggested that AI is on the verge of making the leap from cost-cutting efficiency use cases to true revenue generating opportunities. Again, examples of these are often easy to find in industries with thriving tech relationships and vast amounts of available data, but that can leave the majority of companies wondering why they can't seem to show the same value. Until aging companies are truly willing to commit to changing their process, as well as their tech, to incorporate AI first decision augmentation, we will continue to see these valuable but somewhat boring efficiency-minded use cases in company write-ups.

A.I. in Art and Science

NASA's DART mission to redirect an asteroid uses fully autonomous SMART Nav

Our world's first planetary defense mission was conducted on Sep. 26, 2022. The mission was to determine if the impact of a small space craft could nudge the trajectory of an asteroid. I found the navigation system to be particularly interesting though. Using decades of research from the Applied Physics Laboratory (APL) on missile guidance systems, NASA engineers developed algorithms as a part of the guidance, navigation, and control (GNC) system. It autonomously navigated the space craft 11 million kilometers from earth on solar power traveling 14k mph to a target about 500 ft wide in space, but we are still crashing self-driving cars world-wide. (mission website, photos, live stream)

The Machine Learning community has already moved on from image generation to video. (Meta, Google) What does creative AI mean for the global workforce?

Generator models like Dall-E 2 are taking the art world by storm and are already producing results, like this photo that won its category in a Fine Art competition. Likewise, there are plenty of articles out there debating the value and/or danger of AI generated music, but the machine learning community already had its sights set on a larger prize, video. These new models released by Meta and Google can generate video from text prompts, animate static images, and even enhance existing video with additional graphics. What does this mean for "creative" jobs? In the mid 2000's when some tech start-ups (Spotify, Netflix streaming service) were beginning to have the compute power of the cloud and the vast amounts of data necessary to take an AI first approach to their business models the media erupted with stories about how A.I. would begin to take jobs. All evidence and expertise pointed to a specific timeline in which AI would disrupt the job market. First, highly manual mundane blue-collar jobs like assembly lines would disappear. Next would be the highly repetitive white-collar jobs like data-entry and quality analysis, followed decades later by skilled white-collar workers in high paying positions. Finally, AI would never be able to disrupt the need for skills that were uniquely "human" graphic designers, artists, etc. In reality we are witnessing the exact opposite. AI has struggled to displace the need for hard working data entry professionals and QA experts but has instead created a need for them. In contrast creative jobs seem to be in great peril, closely followed by highly skilled professionals like coders, bankers, and medical professionals. I am still of the opinion that AI will continue to create many more jobs than it threatens but I find it fascinating just how wrong we were on this one.

A.I. in Healthcare

It's time to be honest about the expectations for DeepMind's AlphaFold technology in Drug Discovery

Proteins are complex molecules in living organisms that perform vital tasks needed for all sorts of bodily functions. Molecular biologists can spend years conducting experiments to determine a protein's structure. This is important because the three-dimensional structure gives us an idea of what the protein's purpose is. AlphaFold is a revolutionary advancement in the field of biology that can predict this 3D structure in minutes. Its success has been widely covered in global news media which subsequently theorized its effect on existing businesses, most notably drug discovery. This article summarizes an investigation by MIT that outlines why those claims are little more than conjecture. While this breakthrough should still be celebrated, it highlights the need for professionals in all industries to be more informed on the basics of machine learning, so that unrealistic expectations don't stifle real progress.

There may be a new AI tool to help clinicians diagnose sepsis

I talked with a close friend whose father had slipped and fallen while shoveling snow in their driveway. She described the painstaking process of no less than 12 doctors' visits, several tests, and countless hours in waiting rooms that finally resulted in a 10-day stay in the local hospital. The fall had resulted in several fractured ribs that went untreated and ultimately caused an infection and subsequent blood poisoning as her father's condition slowly worsened. They were lucky in that his hospital stay resulted in treatment, major surgery, and ultimately recovery, but this scenario is far too common in our nation's health care system. The complexity with which sepsis is diagnosed is hard to even describe. It can originate from many different situations and can present in a multitude of ways. As a result, clinicians rely on medical history, physical exams, tests, and clinical knowledge to piece together a diagnosis. Doctors have increasingly adopted technology to aid in detection but the current rules-based (if this then that) engines result in extremely high false positive cases. This is a high-risk, complex problem with clear benefits. We should be able to leverage the technology and data at hand to solve it and save people's lives. Take a look at the article for a deeper dive into those who are already on the case and an examination of the EMR or Electronic Medical Record.


Kaggle's state of Data Science and Machine Learning 2022 (Executive summary, Data)

[Try It Out] Use AI to build power point presentations.

Agility Robotics' Cassie robot breaks speed record

An AI is planning to run for political office in Denmark

Fake Joe Rogan interviews Fake Steve Jobs

Totally Random

I could watch pool trick shots all day

This duck can play the bongos

Biophilic skyscraper is home to 80k plants

Did you know that when you lose weight you breathe it out (seriously!)


bottom of page