“Ken Griffin, founder and CEO of Citadel, said he remains skeptical that artificial intelligence could soon make human jobs obsolete as he sees flaws in machine learning models applied in certain scenarios.” “Speaking to Citadel’s new class of interns in New York, the founder and CEO acknowledged artificial intelligence had reached an inflection point with the rollout of large language models, but disputed that LLM-based tools like OpenAI’s own ChatGPT would over the next three years become more valuable than the top talent he’s recruiting straight out of universities. For a number of reasons, I am not convinced that these models will achieve that type of breakthrough in the near future.” He ”cited self-driving cars, which struggle when confronted with so-called edge cases that fall beyond the boundaries of statistically common traffic occurrences.” They “are trained by feeding them images labeled with data describing the environment around them: That way a computer can recognize for example whether what’s crossing the street is a small child or just a rubber ball. But there are often instances when a self-driving car may find itself confronted with a situation it has never seen before, and it behaves incorrectly since it lacks human intuition.” Furthermore, “Machine learning models do not do well in a world where regimes shift,” which is often the case in the real-world. “Self-driving cars don’t work very well in the North due to snow. When the terrain changes, they have no idea what to do. Machine learning models do much better when there’s consistency.” Unfortunately, the real world is filled with inconsistencies. Many executives ignore these problems because they think they need to talk up “AI. Last month he that “while it has sparked huge investments in data center infrastructure and semiconductor chips like Nvidia’s for AI training, AI’s impact on business remains small. That’s in part because executives haven’t understood how to most effectively harness the power of AI—they just know they need to tell investors like Griffin they do.” “I’ll ask the question, ‘So tell me how you use AI at your firm today,’ and you get these big smiles, and you get these really enthusiastic answers that have almost nothing to do with AI.” Unfortunately, people conflate the contributions from algorithms, Big Data, and AI and end up concluding that AI has made big contributions over the last 10 to 20 years. The title of a recent report by Goldman Sachs says similar things: “GenAI: Too Much Spend, Too Little Benefit?” One partner says: “AI technology is exceptionally expensive, and to justify those costs, the #technology must be able to solve complex problems, which it isn’t designed to do.” Other partners are more optimistic but admit that “AI’s killer application has yet to emerge.” #innovation #startups #hype #artificialintelligence https://lnkd.in/gnSKu4rb
Dr. Jeffrey Funk Thanks for posting - I'd missed this. I must say, I don't agree with Ken. Many are taking the approach to apply what is 3-decade plus mindset to GenAI, looking for how it can be 'poured on top' of existing Business Processes to drive out increased efficiency/effectiveness. And while that is right and needed, it's what is leading people like Ken to reach the conclusion that GenAI/AI is somehow falling 'short' of expectations and/or ROI. My view (and I'm not alone) is that what we're actually seeing is more of a 'Cambrian Explosion' across AI, ML, and NLP - along with other advances in chips, cloud, and network - all creating a J-curve we've not seen prior. But where the 'true' use-case lies is NOT around an organization's Business Processes - it's around its Thought Processes: how it synthesizes, associates, reasons, generalizes, reflects, learns, decides, and - yes - creates. In other words, we - the HUMANS - are the 'killer app'! And the ROI of that is priceless! Note to Klaus Schwab - If the Fourth Industrial Revolution was about 'rapid technological advancement', then the Fifth will be about 'rapid advancement of human cognition'. I believe we are at its gates!
Timothy R.
2w
His statements you quoted show a lack of understanding of the utility of GenAI as an incremental yet economically impactful technology. It (GenAI), of course, will not replace humans. It Augments their work like nothing in history. It's a tool. Augmented Intelligence tech is key to our future. It is also the most important privacy and security tool since the creation of the Internet if used locally I.e open models issolated with in businesses and homes, not propriety models such as OpenAI.
An interesting perspective, however , the automation of warehouses and production plants has slowly impacted those areas , and had a significant impact , already a few of the biggest technological companies have let people go and replaced them with some form of AI while asking people to work with an AI system, that is learning what tasks people perform , yes there are flaws , but not enough to stop the momentum of implementation of AI , keep in mind that the AI systems evolve and grow exponentially, so any errors today will be quickly corrected, the projected savings for a business are significant and will continue to drive innovation and growth at a very quick pace . As I keep saying , we are creating an immortal Alien super intelligence, whatever we may think , a system that evolves exponentially it will not be long before we will see massive changes, I would add that Ray Kurzweil has been accurate in his timeline of events so far .
I see AI as a tool that can improve people's lives, not replacing them.
Jan S.
2w
Completely agree. ML models are very good at solving specific tasks, but bad at generalization. And even if LLMs are perceived as intelligent (false: even if they were capable of passing the Turing test, they probably couldn't do it with Grosz's Extended Turing test: interaction over the long-term in uncertain and dynamic environments), it is more a summary tool plus strong hallucinations... then you ask LLMs how many times a letter is repeated on a word, and they fail. LLMs help generate text from existing knowledge (or doing lazy students' homework), generating boilerplate code, and speeding up the search for what you would have found in Stack Overflow, excepting that, this time, the code is probably buggy/hallucinated, so, you still have to fix it. Some managers try to save money by not training employees, instead, using coding assistants: wrong -and I've seen this. GenAI is good at generating imagery, and similar content. I'm afraid both might be excellent for designing believable scams for cyber-criminals, generating fake AI girlfriends -huge social problems coming (although incels might become less aggressive on social media), etc. But current models, generally, are just helpers, and pretty bad at innovation.
Joe H.
2w
Well, he doesn’t understand decision science, graph theory, & physics as that governs real world scenarios. None of which have much to do with “AI”, but can be coupled with it.
Joseph Urban
2w
Disagree. If a business process can be described, it can be deconstructed (as all procedures should be). That means each step is less complicated than the procedure being described. In my research, LLMs excel when provided sufficient context and narrow scope. Each step (or group of steps) has a reasonable chance of being replaced by an LLM agent or workflow automation. And, LLMs are powerful enough today to define/extract those procedures and even if the experts have trouble explaining them.
Dr. Jeffrey Funk AI 🤖 is the “the thing” that will help you “do something” but AI should not be seen as the lead on anything. If done right AI will most likely be invisible 🫥 while helping things in the background The use cases that will likely be most successful will not be these huge massive high risk and complex problems, they will be solving small problems at scale that allow humans to focus on other things.
Why on earth does AI need to “solve complex problems” in order to be valuable. I need it to 1) Take care of simple repetitive stuff to save me time 2) Bubble up more and more useful information to help me excel . Both are hugely possible now and soon . They won’t replace me nor replace interns , but there is this huge middle ground between “ doing everything always” and “doing nothing good ever” where the future lies. Ken should be reassuring them while also getting the audience to be constantly experimenting with it all and learning where it’s useful and the limitations and the types of changes and developments in how it works and what it does