 |
|
|
|||||
Headlines alternate between massive AI investments and reports of failed deployments. The pattern is consistent across industries: seemingly promising AI projects that work well in testing environments struggle or fail when deployed in real-world conditions. It’s not insufficient computing power, inadequate talent, or immature algorithms. Ive worked with over 250 enterprises deploying visual AIfrom Fortune 10 manufacturers to emerging unicornsand the pattern is unmistakable: the companies that succeed train their models on what actually breaks them, while the ones that fail optimize for what works in controlled environments. The Hidden Economics of AI Failure When Amazon quietly rolled back its “Just Walk Out” technology from most U.S. grocery stores in 2024, the media focused on the obvious: customers were confused, technology wasn’t ready, labor costs weren’t eliminated as promised. But the real lesson was subtler and more valuable. Amazon’s visual AI could accurately identify a shopper picking up a Coke in ideal conditionswell-lit aisles, single shoppers, products in their designated spots. The system failed on the edge cases that define real-world retail: crowded aisles, group shopping, items returned to wrong shelves, inventory that constantly shifts. The core issue wasn’t technological sophisticationit was data strategy. Amazon had trained their models on millions of hours of video, but the wrong millions of hours. They optimized for the common scenarios while underweighting the chaos that drives real-world retail. Amazon continues to refine the technologya strategy that highlights the core challenge with visual AI deployment. The issue wasn’t insufficient computing power or algorithmic sophistication. The models needed more comprehensive training data that captured the full spectrum of customer behaviors, not just the most common scenarios. This is the billion dollar blind spot: Most enterprises are solving the wrong data problem. Focusing on the right data, not just more data Enterprises often assume that simply scaling datacollecting millions more images or video hourswill close the performance gap. But visual AI doesnt fail because of too little data; it fails because of the wrong data. The companies that consistently succeed have learned to curate their datasets with the same rigor they apply to their models. They deliberately seek out and label the hard cases: the scratches that barely register on a part, the rare disease presentation in a medical image, the one-in-a-thousand lighting condition on a production line, or the pedestrian darting out from between parked cars at dusk. These are the cases that break models in deploymentand the cases that separate an adequate system from a production-ready one. This is why data quality is quickly becoming the real competitive advantage in visual AI. Smart companies arent chasing sheer volume; theyre investing in tools to measure, curate, and continuously improve their datasets. First-hand experience As the CEO of a visual AI startupVoxel51these challenges are something Ive lived first-hand. My co-founder and I started the company after seeing how bad data derails AI projects. In 2017, while working with the city of Baltimore to deploy vision systems on its CitiWatch camera network to aid first responders, we experienced the pain of creating datasets, training models, and diagnosing failures without the right tools. That work inspired us to build our own platform, which became FiftyOnenow the most widely adopted open source toolkit for visual AI with more than three million installs. Today, more than 250 enterprises, including Berkshire Grey, Google, Bosch, and Porsche, use it to put data quality at the center of their AI strategy. Here are just a few outcomes: Allstate improved data quality in vehicle damage inspection by automating the pipelinesegmenting parts, detecting damages, and matching repair costsreducing hours of manual effort while ensuring consistent results. Raytheon Technologies Research Center organized and filtered large research datasets to surface meaningful patterns in complex image attributes, turning noisy data into usable insights. A Fortune 500 agriculture tech company curated training data from harvesters to improve grain segmentation, capturing edge cases like unhusked and sprouting kernels for more robust models. A Fortune 500 company curated visual data to detect defective screens before shipment, preventing costly recalls and customer returns. SafelyYou shows the impact of this approach. The companys system helps care delivery in senior care facilities with models that help reduce fall-related ER visits by 80%. The key wasnt just massive scale60 million minutes of videobut the ability to curate variations in how seniors actually fall: different lighting, speeds, body types, and obstacles. By automating checks for annotation mistakes and model blind spots, they cut manual review by 77%, boosted precision scores by 10%, and saved up to 80 developer hours each month. The Path Forward For executives evaluating visual AI investments, the lesson is clear: success is driven not by bigger models or more compute, but by treating data as the foundation. Organizations that prioritize data quality consistently outperform those that focus primarily on technology infrastructure or talent acquisition. Investments in data collection, curation, and management systems are the levers that truly move the needle. By embedding scenario analysis into data strategymodeling how different data quality, diversity, or labeling scenarios impact performancecompanies can anticipate risks, optimize resource allocation, and make more informed AI investments. Ultimately, the most successful visual AI initiatives are those that integrate rigorous data practices with forward-looking scenario planning, ensuring that models deliver reliable performance across a range of real-world conditions.
Category:
E-Commerce
Early in my (Chantals) career, my manager, Scott, shared something in my annual review that Ill never forget. My sarcastic sense of humor made some people uncomfortable. He recommended that I “tone it down a bit.” I felt embarrassed and defensive. Since I was young, Id always leveraged humor to connect and signal mental acuity. The feedback made me question what I thought I knew. Was my presumed superpower actually a liability? The conversation rattled me, and I didnt know what to do with the feedback. So often, early-career professionals enter the workforce and receive technical feedback from managers: fix code this way, prepare for a check-in using this template, sequence slides like this for a presentation. This type of feedback is helpful. Too often though, managers are nervous to share behavioral feedback (like what Scott gave to me). They worry that itll come across as too subjective and therefore not valid or offensive to the receiver. These are reasonable concerns, but unfortunately, perception can impact how your career progresses. It might be jarring (and unfair) to receive this kind of feedback, but you can actually use it to your advantage. If youre lucky enough to have a manager who gives behavioral feedback, heres how to move from unproductively rattled to productively responsive. This way, you can leverage the feedback to grow professionally. Be open (not defensive) As humans, we are wired to self-protect ourselves from danger. Research shows that feedback activates the brains threat response. As a result, it can be difficult to accept feedback. To resist a fight, flee or freeze reaction, start by giving yourself grace. As humans, we all have blind spots. That doesnt mean were not good enough the way we are. Then remind yourself that every piece of feedback is one persons perspective, not a fact. Were allowed to hold it at arms length, examine it, and decide if accepting it would support our professional development. When we are “at choice,” we can treat feedback with curiosity, which encourages growth. Practice gratitude Saying thank you releases dopamine and contributes to overall well-being. This is a great antidote to the “fear of not being good enough,” which we often experience when confronted with difficult feedback. Take a moment to appreciate the thoughtfulness of the person who is trying to help you develop and explicitly thank them. This might sound like, I imagine sharing that feedback was difficult, and Im really grateful you did. Its important I understand how Im experienced by others. Thank you. Ask open-ended questions Resist asking the feedback deliverer for numerous examples to back their point. Remember, its not a litigation. This approach will ensure that you dont receive useful feedback from them in the future. Instead, get curious about their experience of you with follow-up questions like, How did that affect you? What else feels important for me to know? What advice, if any, do you have for me? Resist doing the opposite When we receive difficult feedback, it can be tempting to respond by doing the opposite of what weve been doing. But, critical behavioral feedback we receive is often an overdone strength, not a behavior to abandon entirely. For example, one client, Izzy, exuded optimism. She always saw the best in colleagues or opportunities and could frequently be heard saying, Dont worry, itll all work out! and Sure, its possible, no problem. Unfortunately, over time, her relentless positivity started eroding her reputation. Some people perceived her to be naive and thought that she lacked critical thinking skills. Upon hearing this feedback, Izzy felt self-conscious and began to shift her behavior in a dramatic way. She wanted to prove that she could operate with a skeptical eye, It sounds like I should always be the devils advocate in the room, she said. But this reaction would have created a host of other issues. Other colleagues suddenly saw her as overly negative or even inauthentic. Instead, to support Izzys growth, we worked together to invite a little more critical judgment into her leadership to complement her gift of seeing whats possible. When you get tough feedback, instead of over-dialing, figure out specific behaviors that you might be exaggerating. And then, rather than trying to adjust the dial by 180 degrees, try to change it by just 20 degrees. Make small adjustments How do you adjust just 20 degrees? Experiment with new behaviors. Make the experiments small, easy, and playful so they feel appealing versus daunting. For example, my client, Drew, received feedback that he “talked too much and came off as a know-it-all in meetings.” So he decided to conduct an experiment. For a week, he committed to practicing affirming someone elses idea and asking a curious question when someone contributed in meetings before saying what he thought. This sounded like, “Lisa, I see how that could help progress things. Who else do you think we could involve to make it happen?“ At the end of the week, he reflected on how it went, what he learned, and what he wanted to do more or less of the next week. This type of experimentation enabled incremental growth that led to meaningful shifts in how others saw him. The importance of feedback All of us need to receive feedback to hone and continue to grow our skills. For me (Chantal), I started paying closer attention to the way my humor landed with colleagues. I started noticing when my sarcasm enhanced connection and the times when too much levity diminished psychological safety or signaled less professional behavior. Scotts feedback equipped me to use my superpower more skillfully and navigate the nuanced professional realmwith greater effectiveness. Ultimately, we must all know if our humor isnt landing, our communication is too blunt, or our empathy is overbearing. When we have the courage to hear about how others see us at work and are willing to adjust our behavior, were able to have a bigger impact in our careers and in life.
Category:
E-Commerce
Twenty-five years ago, Google unveiled Adwords, which pledged to enable advertisers to quickly design a flexible program that best fits [their] online marketing goals and budget, Google cofounder Larry Page said at the time. The principle was simple. AdWords allowed advertisers to purchase individualized, affordable keyword-based advertising that appears alongside search results used by hundreds of millions of people every day. That decision was a game changer for Google. Advertising now accounts for around three in every four dollars of revenue the company has made so far this year, growing 10% in the last year alone. The product, since renamed Google Ads, has powered the company to prosperity, cementing its position at the top of the search space. But a quarter of a century on, artificial intelligence could force an overhaul of Google Ads. The shift from traditional search to AI answer engines represents the greatest challenge to Google’s $200 billion monetization engine we’ve ever seen, says Aengus Boyle, vice president of media at VaynerMedia, a strategy and creative agency set up by entrepreneur Gary Vaynerchuk. Thats not because competitors are siphoning away users from Google: The companys global daily active users are up 13% year on year, with nearly 2 billion people logging on to Google services every day, according to Bank of America estimates. But because Google is starting to layer in AI-tailored answers into the front page of its search resultsoften above the advertisements and blue links to sources that helped make its name over the last 25 yearsits ability to bring in ad revenue could take a serious hit. If AI answers start replacing traditional Google searches, thats a real threat to the whole cash engine, says Fergal O’Connor, CEO of Buymedia, an ad platform company. Google makes most of its money from ads tied to clicks. The more queries, the more ad space, the more revenue. The problem is that AI summaries of search results make it less necessary to click through to websites. So far, thats been to the consternation of website owners, who rely on visits to their websites in order to sustain their business models. In time, it could harm Google itself. If people stop clicking through to sites because they get what they need from an AI summary, that entire model takes a hit, OConnor says. Of course, Google will obviously try to wedge ads into the AI answers, notes OConnorand indeed, the company is already doing sobut he says its not a like-for-like comparison. One generative answer replaces a full results page of ad inventory, so its fewer impressions, fewer clicks, and less data flowing through the system, he explains. However, if anyone is best placed to capitalize on those changes, its Google, Boyle predicts. Their clearest advantage lies within Google Adswhich has allowed them to integrate ads into new AI discovery surfaces, like AI Overviews and AI Mode, faster than any of their competitors in the space, he says. OConnor believes that Google will adapt to the new norm, with AI being alteringbut not terminalto the future of advertising. If people genuinely stop Googling and start asking, the whole search economy has to reinvent itself, OConnor says. But if you’ve been around the digital ad space for a few decades, you’ll know that we’ve survived a few events that were billed as being apocalyptic to the industry. Google has had 25 years to understand how best to target and present ads to its users and to squeeze out everything it can from the ad industry. Its best placed to secure another 25 years of dominance, even if it requires some changes.
Category:
E-Commerce
All news |
||||||||||||||||||
|
||||||||||||||||||