Elena Alikhachkina, PhD’s Post

this week in artificial intelligence (though, really, most of this editorial essay from YouTube channel Modern MBA explores why so few of the data-centric startups from the 2010s like Doordash, SoFi, Allbirds, or Wayfair can brag about their operating income; turns out, parsing through all that collected data is prohibitively expensive) ... "The people that worked on Big Data had a vested interest in keeping the technology trendy despite the lack of results, and are now doing the same with AI. Whether it's bottoms-up or top-down adoption, money talks, even for those who are simply implementing the technology itself. This is why there's never any end to the online debates between React versus Angular, Kubernetes versus ECS, and so on. You can see that same tribalism in AI where there's no consensus on what tool is best, Tensorflow or PyTorch, or which model is most accurate. The premise of Big Data was that you could unearth hidden innovation, business value, customer insights, and market patterns from data that was simply too overwhelming for a human to digest and analyze. These days everyone has seemingly forgotten about Big Data's failed promises. But the current premise of AI is even more confusing. The new narrative is that data is inherently too complex. Instead what we're supposed to believe is that all that promised innovation, buried insights, and hidden business value is still in this data, it's just that humans can't pull it out. Instead, our only solution is to trust these artificial models—where only a few people have a true understanding of what's really happening under the hood—for perspective and insights; we should be told the answers and not seek it ourselves, and we should value the digestibility and presentation above the accuracy of information itself. For Big Data, every company had its own data, timelines, politics, and priorities, which made execution completely unique—every implementation was a Snowflake deployment. Since no one has achieved any value with Big Data, no one really knows, even now, nearly a decade later, how to actually implement and derive value. No one can authoritatively say: This is how you execute, and this is how you can replicate step for step what we did for your own business. Engineers essentially are reinventing the wheel but never finishing."

🤖 #AI #News You gotta love the battle to control the AI narrative in the news and headlines over the last 12-18 months. It's a constant back & forth between several factions vying for what is the truth of AI, the promise, the dangers and the return on investment. And like almost everything in life, it's a big "it depends" question- but more so "it depends on what you mean by x, y, z" more than it does "it depends on what is the most probably outcome". The following article got a few AI podcasters I listen to a bit testy, and while I think they make some good rebuttals, it more than showed the biases that obviously exist in the sector today. This article in the The Economic Times titled "Goldman Sachs puts a $1 trillion question mark over Generative AI" brings the question of whether AI is all or mostly hype to the fore, and quotes experts saying it's not going to be as impactful as everyone expects: "Acemoglu estimates that only a quarter of AI exposed tasks will be cost-effective to automate within the next 10 years, implying that AI will impact less than 5% of all tasks." I'm not a deep researcher of AI economics (or any economics for that matter) so I'm going to have to accept these figures at face value, but it does interest me a lot on what face of AI we're all looking at this thing that's captured the industry so significantly. I'm sure I'll have a lot more thoughts to throw into my newsletter (https://lnkd.in/gyacn8tB) when I get around to it but one comment that did stand out to me from the article, not as AI specific, but as a tech industry specific pitfall, and that's: "This risk seems particularly high today for using AI to advance automation. Too much automation too soon could create bottlenecks and other problems for firms that no longer have the flexibility and trouble-shooting capabilities that human capital provides." In my experience in engineering, you want to figure things out first (as much as your deadlines or task-masters will allow of course) and then automate that. Just like the adage in Data science "garbage in = garbage out" (don't quote me on this), in systems engineering if your thing is crap, all you're going to automate is crap. Don't automate crap. Have a read of the report for yourself: https://lnkd.in/gp9a_Kn9 (it's long) ET Article: https://lnkd.in/grS_8KKT #AI #AIEthics #AIForGood #DeepLearning #MachineLearning #ArtificialIntelligence

The Nine Truths outlined by Marinela Profi are right on! A good read. The three biggest … Truth #1: Large language models alone do not solve business problems Truth #2: LLMs cannot be directly used for analytics and decision intelligence. Truth #9: If you think your competitors have GenAI all figured out, think again. In net, future-state #AI in business will take on a highly architectural approach that fuses GenAI, LLMs, SLMs, decision intelligence and causal AI to power networks of AI agents that collaborate to help people solve problems and act more autonomously, but with accountability. #AgenticAI #GenerativeAI #CausalAI #SLMs #LLMs #DecisionIntelligence

Generative AI is a long-term investment, not a quest for single-shot success stories. 💡 As we enter Q4 2024, I have collected some lessons learned so far. If you are a decision-maker looking to invest in generative AI, you might want to take a read before jumping in. Are there other lessons/truths you would add to the list? 👇 #generativeAI #artificialintelligence #datamanagement #datagovernance #dataquality #dataengineering

So true! The biggest hurdle I encounter when taking AI use cases with our clients is data quality. Securing the right data governance structure around key data assets is key to harvest the power of AI.

The Economist is wrong about AI... Now, what the clueless editors at The Economist really mean is that GenAI, and specifically LLMs of all kinds, are losing their hype. You see, these folks couldn’t tell a regression model from a Monte Carlo simulation, so to them, it’s all just mysterious “AI” like something out of a Spielberg movie. “Look, Ma, it can talk!” But that’s not the point. You can’t expect much understanding of tech from English majors or the ability to tell the broader AI field from GenAI anyways. What’s interesting are the reasons now being cited for why GenAI hopes, specifically, have not panned out, especially when it comes to areas that require correctness and precision. Some say, “Oh, well, it wasn’t really a good fit for some of the use cases.” Or better yet (or worse), they blame the lack of employee training for some of the GenAI failures to impress. But what really happened is that the reality of LLMs as stochastic generative engines, which can only do contextual reasoning at best, started biting many efforts in the rear once they were put into production. Everything looked so nice and wonderful in demos presented to the C-suite and various boards. But once GenAI started being utilized for real-world use cases, with outliers hitting them left, right, and center, the deficiencies of the LLMs really made many pause. It turns out that when your LLM is downstream from all of your fancy RAG infrastructure, it can still hallucinate like a hippie on LSD. It turns out that the supposed guardrails are not as bulletproof as the demos would have you think. It turns out that the LLMs cannot deal with the many outliers and low-probability events the way a human can, and letting it run amok in a fully autonomous fashion is like setting a time bomb in your enterprise workflows that can go horribly wrong at any minute, with few real controls to stop it. It turns out the C-suite management may be personally liable for things going haywire, and the risks of deploying unproven and hyped-up tech in mission-critical contexts in production can have real consequences for the boards, execs, brand loyalty, customer satisfaction, and most importantly, the bottom line. The silver lining is that there is a lot more to AI than the monkeying and hype of GenAI and LLMs specifically. There is work happening under the radar that doesn’t receive as much attention, with far more promising outcomes than LLMs could ever hope to achieve. Though they will probably take their place in the AI ecosystems of the future as NLP interfaces, once we realize that stochastic methods of sequence generation need to be abandoned in lieu of greedier decoding methods that can actually provide deterministic outputs to the same inputs and hence can be tested and relied on. #ai #genai #bubble

✦ GenAI faces a DATA OVERLOAD crisis, urging a focus on SPECIFIC GOALS for AI implementation, as emphasized during TechCrunch Disrupt 2024. ✦ Prioritizing PRODUCT-MARKET FIT over scale is essential in AI development, with an emphasis on DATA QUALITY, real-time data, and starting small. ✦ Human INVOLVEMENT in AI applications is crucial for successfully addressing specific goals. #GenAI #AIIntegration #DataQuality #TechCrunchDisrupt

Hmmm, maybe we should start the discussion with 𝗮𝗿𝗲 𝗗𝗲𝗲𝗽 𝗡𝗲𝘁𝘀 𝗿𝗲𝗮𝗹𝗹𝘆 𝗶𝗺𝗽𝗹𝗲𝗺𝗲𝗻𝘁𝗶𝗻𝗴 𝗮𝗻𝘆 𝗔𝗜 or are they just pre-processing units organizing the information for a (still missing) brain=#AI? At this point, we implemented the input modalities and the output modalities (semantic #labeling/#NLP and #generators). We still missed AI, but nobody seems to notice or care? We use the brain of the user to understand the decoded input and to initiate the generation of the output. The #intelligence is in the user for now. This would also explain my current problem with "AI robotics". Connecting input to output (perception to actuators) is a fast responding unit responsible for #reflexes. It just executes pre-planned moves to be fast. We need the still missing #brain to do real #planning and #decisions (in all AI aspects but especially in #robotics).

As someone deeply fascinated by the advancements in AI, particularly in the realm of Generative AI (Gen AI), I believe it's crucial to address a fundamental aspect often overshadowed by the buzz around larger language models (LLMs). When connecting these cutting-edge AI models to real business use cases, we mustn't overlook the critical importance of data quality, quantity, reliability, completeness, and security. These factors are the bedrock upon which businesses can harness the true value of AI versus its cost. Imagine the possibilities of having Integrated Business Planning (IBP) systems as our true north star. To achieve this, businesses need faster, accurate, and better access to key data streams: 1️⃣ 𝐃𝐞𝐦𝐚𝐧𝐝 𝐃𝐚𝐭𝐚: Insights from commercial organizations are invaluable. From marketing innovations and consumer trends to media spend and performance analytics, understanding demand dynamics is essential. Sales data, covering pricing strategies, promotions, distribution channels, and even the quality of retail space, holds immense significance, particularly for manufacturers. 2️⃣ 𝐒𝐮𝐩𝐩𝐥𝐲 𝐃𝐚𝐭𝐚: The backbone of any business lies in its supply chain. Access to supply data, including inventory management and point-of-sale systems, empowers organizations to optimize operations and respond effectively to market demands. 3️⃣ 𝐌𝐚𝐜𝐫𝐨-𝐄𝐜𝐨𝐧𝐨𝐦𝐢𝐜 𝐃𝐚𝐭𝐚: Integrating macro-economic indicators provides a broader context for decision-making. Factors such as GDP growth, inflation rates, and industry trends offer valuable insights into market conditions and future projections. By prioritizing the quality, reliability, and security of these data inputs, businesses can unlock the full potential of AI-driven insights. Integrated Business Planning systems, fueled by high-quality data and powered by AI, have the potential to revolutionize how organizations forecast demand, optimize supply chains, and navigate dynamic market landscapes. Let's remember: while AI models like Gen AI offer incredible possibilities, it's the quality of data that ultimately determines their efficacy in driving business outcomes. Here's to embracing data excellence as the cornerstone of AI-powered innovation! #pepsicoproud #AI #GenAI #DataQuality #IntegratedPlanning #BusinessIntelligence #Innovation

When it comes to AI, data deserves our unending attention. We cannot be complacent with it, nor grow tired of it. Here's why. https://lnkd.in/eEgQYzEz Written by Ayman Sayed of BMC Software