Carole Hussey’s Post

As Google I/O 2024, with it's breathless coverage of the world-changing AI that's here right now, comes to an end today, I found this article to be very timely. Press Pause on the Silicon Valley Hype Machine https://lnkd.in/eARCNrrz

A scathing, but realistic, take on AI from the New York Times today. A good reminder of how people outside the tech bubble view AI, and it's backed up by facts. I'd be so much more excited about advances in AI if people would just be more honest about its limitations and costs. Worth a quick read, here's a gift link:

This is a good article with a lot of solid evidence to back up its take that AI may be "too stupid and unreliable to be useful." I personally often think about how recent advances in AI were likely driven more by massive increases in training data than improvements in algorithms. Perhaps the reason we haven't seen any more big leaps is that we don't have ten more internets to scrape for the next 10x improvement, and there are serious questions about whether all the data that *has* been used to train current models even should have been. I think AI has its uses, but the fervor with which people pour billions of dollars into it feels more like religious dedication to an imagined sci-fi future than a realistic cost-benefit assessment.

Might the AI hype bubble burst? This article is more conservative than my posture, but there is a realism here often missing in conversation. Many limitations remain on generative AI when we consider professional, reliable applications. ROI is a requirement, not a nice to have feature. More performance lift than cost. Some power alley applications for genAI get too little attention, in my view, in the quest for question answering solutions that are the most challenging to render. Editing, translation, summarization are really high performing and worthy of attention. With mitigations, these can save a lot of time. The costs are real. The value must be engineered. Recognize concepts for what they are and focus on deploying tested, reliable capability. https://lnkd.in/gThQmT_X

I broadly agree with this argument, but it also misses how #generativeAI unlocks new areas for businesses. To give an example I've seen with Legislata. Transcribing audio costs around $100/hour with traditional methods. If you want to transcribe a 5-hour long city council meeting, that's cost-prohibitive. There's simply no market that would pay for that - certainly not for smaller towns or for subcommittee meetings. But with AI speech-to-text models, we can produce a reasonable transcript for 1/1,000th of the price. At that price point, there is a market, and now people can have access to a useful transcript that previously didn't exist. From there better analysis, more informed debate, and faster research is possible. The hype bubble coming from Silicon Valley is definitely a headwind for anyone working in this space. Potential users are inherently skeptical of any claims. But there is definitely something amazing possible with these tools. https://lnkd.in/esr5vw52

Firstly, a big shout out to J. Walker Smith for distributing this article, which is REALLY worth a read. I was compiling a second article yesterday, summarising some of the key findings from the Data & Insights Network Community on A.I. (an excellent and very insightful initiative from the Dutch association), which reinforce/echo most of what Julia Angwin sets down in the attached article. The Dutch are a very digitally literate society, and are keen advocates for the benefits that A.I. can bring to the workplace, but they are equally strong advocates for testing, testing and re-testing the applications to ensure they do not "hallucinate" (what a GREAT word for "mistake"!!), and are insistent upon human oversight. Those that are heavily invested in A.I. will no doubt categorise articles such as this as "Nay-saying", but for us to be able to jump into the future, we must have one foot firmly planted in the present...content is King, but Context is God! #ai #marketresearch #insight #mrx #esomar #insightsassociation #mrs Judith Passingham Melanie Courtright Debrah Harding Jerome Sopocko Dan Quirk Simon Chadwick Philippe Guilbert Frederic-Charles Petit Phil Sutcliffe Lucy Davison Crispin Beale Danny Russell David VL Smith Wim van Slooten Theo van der Steen Henk Noort Jean-Marc Leger Ian Kiernan Charlie Butler David Durnford Dr Nick Baker Paul Kingsley-Smith Ben Page Jon Puleston Martin Oxley Andrea Rademeyer Rhiannon Bryant Daan Versteeg Reineke Reitsma Robert Schueler Jaap Wieringa

A great article from The New York Times and I completely agree with the commentary from J. Walker Smith. You cannot get away from the AI hype this year. It is a nice distraction from the disastrous economy/pending election chatter though. I am dabbling with it for my communications projects and it is useful when I start to plan content. It maybe saves me a little bit of the online research I'd start with. But if it comes up with anything substantive, I then do need to fact check so that adds time. The grammar and writing tools help me think about things differently, but they're not doing my job for me. I'll take what they give me and reshape it, but it certainly isn't giving me exactly what I need. The image generator tools I do love though. Being able to create something different or tweak what already exists to reflect specific branding elements is great. I also saw a tool demonstrated the other day that can automatically resize an image for different uses, a really useful timesaver on a task that's needed but isn't creative. So for me, the value of AI at the moment is: ✔ Starting to shape initial ideas ✔ Helping to think of alternative ways to express something ✔ A final check to see if what I've created could be improved It isn't truly creative, and sometimes it's totally off beam, but seeing something you don't like can often bring you closer to what you do. I think that some of the challenge is asking questions in the right way. The downside of AI is you can go down a wormhole just because of the hype and you think you should be using it. The danger being you don't actually save any time or produce anything better. #AIforCommunications #ContentCreationAI

Interesting take on AI. We can disagree with the interpretation but the facts seem beyond dispute. Rates as a “must read” in my view. Very well written too.

Last week was yet another big one for AI. On Monday, OpenAI released their GPT-4o, promising a more chatty, emotional experience, as well as a general upgrade (though not as big as GPT-3 to GPT-4 or the promised bump to GPT-5). Maybe most intriguing, it was also released for free. Next came Google’s I/O, where they unveiled an AI-powered search engine, AI helpers across their products, and their own answer to GPT-4o, Project Astra. And right on cue, Nvidia is expected to release earnings this week, which will either keep their $2.3 trillion dollar valuation (+240% last year) rising, or signal a warning. Against all of this, I want to consider a New York Times editorial from last Wednesday, by the outstanding investigative reporter Julia Angwin. She suggests not only that AI is extremely over-hyped, but mocks the call to put a six-month pause on new development last year, and debunks many claims by AI developers (such as ChatGPT’s performance on the bar exam and Google’s claims that AI discovered millions of new chemical compounds). She compares the technology to the Roomba vacuum, and asks us: “The biggest question raised by a future populated by unexceptional A.I., however, is existential. Should we as a society be investing tens of billions of dollars, our precious electricity that could be used toward moving away from fossil fuels, and a generation of the brightest math and science minds on incremental improvements in mediocre email writing?” As a CEO in the tech industry, I find the collision of these viewpoints exciting, and highly relevant to consider. As we see stocks breaking new barriers, we must balance with skeptics who predict the AI bubble will soon burst. It allows us to pair giddy excitement with rational pragmatism. My question is this: in one year, do you think this take will be viewed as uncommon wisdom, or, should AI continue its rapid surge forward, short-sighted and over-stated?   https://lnkd.in/gs-nXn7Y

I couldn't agree less with this article. In my view, AI will change the world as much as, if not more than the internet did...which is a lot. '60s computer scientist Roy Amara said it best: “we overestimate the impact of technology in the short-term and underestimate the effect in the long run.” Both Altman and the author are falling victim to this blind spot, but to me, it seems far more dangerous to underestimate the impact and get blindsided by the change than it is to be slightly overzealous on estimating the rate of change. #AI #ArtificialIntelligence #Technology #Innovation #cdntech [Read more](https://lnkd.in/gJWwyFZR)