Gianni Giacomelli

🤚🏼 🛑 How to identify LLM use cases? Prof. Thomas Malone, who leads MIT Center for Collective Intelligence, has developed a framework to help us identify areas where hashtag #LLMs can shine. Here's how the "Collective Brain" model breaks down LLMs into six key areas: 1️⃣ Sense LLMs can analyze massive datasets to find trends, knowledge gaps, and biases. This can improve decision-making and transparency in supply chains or customer interactions. 2️⃣ Remember Train LLMs on your company's data to remember and retrieve information effectively. Think customer service procedures, sales pitches, or employee policies. 3️⃣ Create LLMs are whizzes at generating creative text formats. Use them for content creation, bioengineering or mechanical design, or even brainstorming with tools like MIT's Supermind Ideator. 4️⃣ Decide LLMs can be your decision-making partner, analyzing data, suggesting options, and highlighting potential risks. Valuable for managers, healthcare workers, policymakers, and more. 5️⃣ Act While LLMs can't directly control physical processes, they can guide human actions or lower-level AI tools for execution. 6️⃣ Learn LLMs continuously improve by analyzing the outcomes of actions and comparing them to expectations. Human feedback and reinforcement learning are key here. 🤔 How prioritize those use cases? A few criteria are emerging: ▶ How important is accuracy? ▶ How important is the speed of execution? ▶ Are there responsible practices available? ▶ Can we quality-control? And what about you? How are you identifying LLM use cases within your organization? I’m reading your comments! ✍ ✍ ✍ #atmira #shaperbyatmira #smarteconomy #artificialintelligence #Innovation #AIinEverydayLife #TechInnovation #DigitalTransformation #businesssolution #ai 𝗟𝗶𝗸𝗲𝗱 this post? 𝗙𝗼𝗹𝗹𝗼𝘄 𝗺𝗲 for regular insights. 𝗛𝗶𝘁 𝘁𝗵𝗲 𝗯𝗲𝗹𝗹 icon 🔔 on my profile to be notified when I post. әʌ!ʇɔәdsɹәd ɹәɥʇouɐ ɯoɹɟ ʎʇ!|ɐәɹ ɹnoʎ әәs noʎ d|әɥ ! '!ɐ puɐ s!dɐ ƃu!sn Picture credits: MIT Center for Collective Intelligence

11

Excellent DEI in Tech insights that is a must read for all leaders in Tech.

13

2 Kommentare

Generative AI is changing the enterprise in five significant ways: shifting our relationship with data for decision-making, reconfiguring the supplier ecosystem, driving the reimagination of processes, reshaping the skills enterprises need to succeed, and blowing up how our companies are organized and how we must manage teams. Enterprises must react and prepare against the emerging threat of 'born GenAi' businesses, just as markets were disrupted by those 'born Digital'. To do so firms must address all five of these changes as they race to take advantage of the benefits of becoming a Generative Enterprise™. Dana Daher and I dive deep. HFS Research Tony Filippone Elena Christopher Phil Fersht Saurabh Gupta Joel M. Thomas Reuner

24

2 Kommentare

Super interesting seeing #GoogleIO with #Gemini today vs #OpenAi and #Gpt4o yesterday. https://lnkd.in/dNXGBwvA Multi-modal definitely the hot space. OpenAi launched. Google showed a pilot. Both focusing on voice and video as the new interface. Gemini reviewing code from live video is mind blowing. Education clearly a focus - Google showing voice interaction on the back of NoteLM Huge differentiator for Google is their context window - they announced consumer availability of their 1m token context window and expanded the Google Gemini Pro context window to 2m tokens. Honestly, in my experience, this completely differentiates Google from the competition. Some mention of Gemini agents. Agentive workflow is IMHO the biggest revolution ahead. Gemini agentic workflows to offer consumers easy return of purchases etc will be a game changer. More example in IO than launch. Clearly Google is focusing on incentivising consumers to stay on their platform and interact more to fight any flight to OpenAI, Microsoft and Meta. So lots about Gemini integration with Google Photos and Workspace. New evolution of Google search with more Gemini integration. Generative capability for video, photos and music announced - these look mind blowing. Google showed lots of Gemini integration into Gmail and Workspace - personalized AI integration. Also a prototype of an AI Gemini team assistant building a collective memory of team interaction. IMHO, this is Google launching Retrieval Augmented Generation (RAG) and agentic workflow to keep users on platform. Google announced Gemini Advanced with planning. Planning has been highlighted as the Achilles heel of current AI. This will be a big play for gmGoogle leveraging their view of your personal data. There is the Gemini app too - like ChatGPT. Again natively multi-modal. Personalized AI assistants. But Google launched the equivalent of custom Gpts called Gems. Then probably Google's advantage - the Android install base. Announcements of coming changes with AI at the core (you can already replace Google Assistant with Gemini). They also announced "Circle the search" - effectively allowing you search based on circling on screen. This also launching immediately. Gemini integration on the phone is also fully context aware (the open document, etc). The announcement of Gemini Nano with multi-modality coming to Android is AI / LLM coming to device. This is the game changer for responsiveness, privacy and poor data connectivity in remote / poor locations. This is where Google can trump Open AI and leads Apple's lagging AI. Both Google and OpenAI showing massive incentives for consumers to provide huge amounts of Multi-modal training data. It's unclear if Google was more cautious vs OpenAI jumping the gun on Multi-modal - ChatGPT severs appeared to struggle to keep up with audio since last night. Lots of what Google announced were pilots and gradual roll outs.

1

2 Kommentare

A significant amount of Boards of large corporations display a shortage of technology expertise, which potentially affects their competitiveness in today's rapidly evolving landscape. The recent report, ‘Bending Time’ in the Age of Generative AI: Is Your Board Ready?, co-authored by #Accenture's Arjun Bedi and Selen Karaca-Griffin, along with #Fred Hassan , explores strategies to address this crucial gap and suggests three ways companies can embed technology expertise in their boards. Learn more: https://lnkd.in/dchvDph9  

18

Boards face global challenges, but one key unites them: Adaptability. Our report finds economic uncertainty, geopolitical shifts & the need for transformation are top of mind for boards worldwide. From talent to tech, boards must be strategic and lead companies through uncertainty. Explore the three area reports for an all-encompassing perspective: #BoardMatters #BetterWorkingWorld

14

#GenerativeAI could significantly change how companies operate- 50% of CEOs expect #GenAI to enhance their ability to build #trust with #stakeholders & abt 60% expect it to improve #product or #servicequality- Key #insights from PwC 2024 #CEOSurvey ✅ 45% of global CEOs believed their companies would no longer be viable in ten years’ time if they continued on their current path. ✅ The impetus to reinvent is intensifying. CEOs expect more pressure over the next three years than they experienced over the previous five from #technology, #climatechange and nearly every other megatrend affecting global #business. ✅ CEOs perceive enormous #inefficiencies across a range of their companies’ routine activities—everything from #decisionmaking meetings to emails—viewing roughly 40% of the time spent on these tasks as inefficient costing US$10 trillion tax on #productivity. Generative AI, which about 60% of CEOs expect to create efficiency benefits, could help relieve some routine burdens. ✅ Four in ten CEOs report that they have accepted lower hurdle rates for #climate-friendly #investments than for other investments—in the majority of cases, between one and four percentage points lower. This is clear evidence that some CEOs are willing to make complex trade-offs as they strive to boost the #sustainability of their businesses. ✅ Two-thirds of CEOs report a reallocation of #resources (#financial and #human) of 20% or less year to year. ✅ One in four anticipate reducing #headcount by 5% or more in 2024 due to generative AI ✅ When it comes to generative AI risks, CEOs are most concerned about #cybersecurity #AI #FinTech #RegTech #Bigdata #dataPrivacy #Infosec #Futureofwork #FinServ #Regulation

4

All (for profit) companies exist with one simple objective: Maximize shareholder value. We would love for them to focus on things like ESG and DEI initiatives too - stakeholder value - and companies will do that if it drives objective #1 above. So let's talk about shareholder value in the channel. (This post was inspired by Insight today). The lowest valuations in the $4.94 trillion tech industry are carried by distribution. With a business model primarily built around transactions, supply chain, and little in terms of end-customer relationships, investors respond accordingly. For example, the largest distributor in the world, TD SYNNEX is valued at 17 cents to the dollar (You could acquire $56 billion in revenue - more than Coca-Cola, McDonald's, or Nike - for slightly less than $10 billion). The next-lowest valuation is pure resell revenue. Also transactionally focused, the value comes in the end-customer relationships and higher probability of repeatability. One of the fastest growing resellers in the world, Computacenter, is valued at 42 cents for every dollar of revenue. Today, Insight acquired Infocenter who was ServiceNow's partner of the year. This came a few months after acquiring SADA, An Insight company, who was Google's partner of the year 6 times(!) Under the leadership of Joyce Mullen, a long-time channel luminary, Insight has been diversifying its revenue and investing in growth markets. This has led to a near-doubling of the stock price in 2 years and a current value of 63 cents to the dollar ($9.2 billion in revenue is valued at $5.8 billion). Interestingly, SADA and Infocenter were both growing much faster than the vendor's they specialized on. To put it in perspective, Google is worth $6.50 per dollar of revenue and ServiceNow a staggering $14 per dollar. Partners can grow their business faster than vendors because of the multiplier effect - the amount of services a customer requires to get the products to work is up in the 6-7X range of what they spend on the products themselves (whether done in house or outsourced to a third party). Services as whole across our industry (today make up 2/3 of the entire pie) are also outgrowing most hardware and software categories. Should partners who are specializing in fast growth markets and outgrowing some of the titans of industry be valued higher? World GDP growth this year will be shy of 3% while the tech industry will double that (6.2%). Should partners such as managed services providers who are growing at double that (12%) be worth even more than today? Does winning partner of the year inside a rocket-ship platform ensure you get into the M&A room at the major players? Thoughts?

150

13 Kommentare

Focusing on a better future is a powerful mind-hack to keep us going -and improving- during adversity. In the same way, organizational resilience is not a survival tactic—it's the path to accomplish and sustain competitive advantage. Our study shows which organizations that understand resilience this way, and prioritize it, are 1.7 times more likely to achieve and sustain profitable growth compared to their peers. What strategies can leaders employ to grow resilience at the service of profitable growth? In this very data intense report we explore the approaches that set apart resilient companies from the rest and we provide executives with the tools to grow its financial return. Now and in the future. Thanks to my co-authors Muqsit Ashraf | Rachel Barton | Miguel Torreira | Ladan Davarzani and to my amazing research team for putting together the incredibly rich economic and data science analysis that support these strategic insights Ana Ruiz Hernanz | Sebastián Bonals | Julia Segoviano | Lucia Pezzarini | Maria Victoria Arbeletche | Ramiro V. Kossacoff | Juan Pablo Romero https://lnkd.in/dXSMtdZJ

38

2 Kommentare

Some personal reflections on the discussions and thoughts that have been shared on yesterdays Meetup "Data Mesh: Providing Value with Data Products" which I attented with my colleagues Fabian Wilckens and Jonathan Orschiedt. 🙌 1️⃣ The concept of #dataproducts is here to stay since there is no other alternative on the horizon, yet. Period. 2️⃣ #Datamesh is not prerequisite to introduce the data product approach to an organization. The #dataproduct approach is universal and also works in other organizational setups such as hub-and-spoke, etc. 3️⃣ There is no one definition for a #dataproduct, yet companies should rigorously define a data product or types of data products for the underlying organization in order to build a common sense, shared understanding and therefore consistency. Hint: There are existing definitions on types of data products which can be built upon. 4️⃣ The role of a #dataproductmanager is a very young discipline and organizations have to define the role porperly and to their needs by defining accountabilities and responsibilities. On a personal note: Communication is a key skill. 5️⃣ #aiproducts are #dataproducts - maybe currently very hyped - yet they are #dataproducts. At Mindfuel we are pioneering #dataproductmanagement with our customers, drop a dm if you are interested in an exchange, also happy to read on your reflections in the comments 👇 #dataproductmanagement #dataproducts #aiproducts

7

3 Kommentare

#Predictive text - The power and usefulness of #AI, #LLM, #MachineLearning, #DeepLearning Question for reader - what is #Pinyin: Pinyin is the phonetic writing system of the #Chinese #language “#Hanyu Pinyin, or simply pinyin, is the most common #romanization system for Standard Chinese. In official documents, it is referred to as the Chinese #Phonetic #Alphabet. It is the official system used in #China, #Singapore, #Taiwan, and by the #UnitedNations. Its use has become common when transliterating Standard Chinese mostly regardless of region, though it is less #ubiquitous in #Taiwan. It is used to teach Standard Chinese, normally written with Chinese characters, to students already familiar with the Latin alphabet. The system makes use of #diacritics to indicate the #four #tones found in Standard Chinese, though these are often omitted in various contexts, such as when spelling Chinese names in non-Chinese texts, or when writing non-Chinese words in Chinese-language texts. Pinyin is also used by various input methods on computers and to categorize entries in some Chinese dictionaries. Hànyǔ (汉语; 漢語) literally means 'Han language'—meaning, the Chinese language—while pinyin literally means 'spelled sounds'” #wikipedia for Pinyin:

Citation from our latest paper "Data products, data mesh, and data fabric New paradigm(s) for data and analytics?" with Ivo Blohm, Felix Wortmann and Christine Legner: "Data products, data mesh, and data fabric reflect comple- mentary concepts to foster data-driven innovation at scale in organizations. The concepts are promising to make data usage more effective, agile, and scalable by adopting an enterprise-wide perspective and emphasizing socio-tech- nical principles." (Blohm et al. 2024) If you are interested in #dataproductmanagement, please find the paper here for further reading: ➡ https://lnkd.in/dz4SRBMd and want to exchange on the #dataproduct concept, please feel free to book a meeting: ➡ https://lnkd.in/dduGgGjD #dataproducts #dataproductmanagement #datafabric #datamesh

6

Consumers Know More About AI Than Business Leaders Think by @BCG Business Leaders should understand (and not underestimate) consumers when developing and deploying AI-enabled solutions and transformations! #rafAIgual #GenAI #Consumer #solutions #leadership

4

🎓 The Foundation Model Transparency Index. #Transparency is one of the pillar of the #ResponsibleAI. At the Stanford University there is a dedicated team, the #CRFM (Center for Research of Foundation Models) which is focused on defining and regularly update the transparency foundation models index of: "How is transparency best measured, how should companies disclose the relevant information publicly, how should scores be computed/presented, and how should findings be communicated to companies, policymakers, and the public? The Index aims to measure transparency to bring about greater transparency in the foundation model ecosystem: the board's collective wisdom will guide the Index team in achieving these goals". The #May2024 Foundation Model Transparency Index was created by a group of 8 AI researchers from #Stanford University's Center for Research on Foundation Models (CRFM) and Institute on Human-Centered Artificial Intelligence (#HAI), #MIT Media Lab, and #Princeton University's Center for Information Technology Policy: 📌 Website of the CRFM: https://lnkd.in/ddjm4kpx 📌 The May 2024 FMTI results: https://lnkd.in/dmysJ6s2 📌 The May 2024 FMTI Paper (also embedded): https://lnkd.in/d-QGDnCE ✅ Key Findings Summary: ✔️ The average transparency score is 58, with the highest score being 85 out of 100. This marks a 21-point improvement in the mean score compared to the October 2023 FMTI; ✔️ Improvement overview: Significant improvements were noted across the board: the top score increased by 31 points and the bottom score by 21 points. All eight developers assessed in both the October 2023 and May 2024 FMTI have improved their scores; ✔️ Indicator satisfaction: Out of 100 transparency indicators, 96 are met by at least one developer, and 89 are satisfied by multiple developers; ✔️ Proactive disclosure: Developers now proactively release transparency reports, a shift from the previous method where the FMTI team collected information from the Internet. On average, developers disclosed 17 new indicators-worth of information in their reports. ⚠️ The consistent improvement across all developers assessed in both periods highlights a positive trend towards greater transparency in the foundation model development community but there is still a gap for improvement with a mean score of 58 out of 100. 💡In the #NextSteps the researches highlighted a remarkable point which I believe should be not a desiderata: ✔️ "#Policymakers: by identifying areas of sustained opacity (no improvement from October 2023 to May 2024) and systemic opacity (all companies disclose little or no information), we make clear where policy intervention may be necessary for transparency. By specifying precise indicators, we demonstrate how policymakers can define clear disclosure requirements." 🔍Check the comments for more content. #AI #LLM #Models #Research #Academy #ArtificialIntelligence #IA

10

3 Kommentare

Innovation needs convictions! How to develop a conviction? Problems, Insights and Hypothesis. Convictions are formed by opinions, opinions are born from hypothesis, hypothesis are drawn from observations and insights... "Sorry we only take cash": ever encountered one of these situations?  That used to drive me absolutely mad (and it still does although in fairness most businesses tend accept cards now). Why? Because: - I don't like to count change, carry coins and bills,  - It's slow an inefficient and adds to queues at cash out - I like to track my spending - The first credit card was invented in 1950 (!), the Diners Club, so by no means using plastic to pay is new...! Frustrations can be a great source of innovation. Frustrations are emotions, but they can get converted into insights and observations. In my case, my frustration led me to work on designing and deploying contactless payment systems across transport and retail on 3 continents for close to ten years! Such a great opportunity to bring convenience to millions of daily commuters and shoppers! How to bring a conviction to life? Leadership, Followers and Execution Leaders need followers to help them achieve their grand plans (most of the time ruthlessly), so they can achieve their goals. They inspire through their charisma or their execution skills to sell their vision to others. That's how they get investors, customers and loyal employees to support their journeys. Amazon's founder Jeff Bezos created the second largest retailer in the world but he was flipping burgers at McDonald's when he was 16. Today, 1 out of every 153 American workers is an Amazon employee! How to allow convictions to emerge in an organization? Frame, Space, and Support. For years, Google used to have the 20% rule for their employees to spend 80% of their time on their job, and 20% on projects of their choice, unrelated to their regular job responsibilities; giving birth to products like Gmail and Google Maps. The lesson is it's not enough to just encourage new ideas. They can only develop and mature into convictions and morph into reality with a framework to institutionalize the creative process end to end. Read the full story here: https://lnkd.in/g4FkiEjQ How about you? What convictions led you to where you are now? #innovation #technology #personalgrowth #koncentrik __________________ P.S. for more of these check out ‼️ https://koncentrik.co ‼️, THE newsletter you need to follow Technology-led Innovations. For tech builders and business leaders.

13

-The first tier is hypostasis — a credence or values system, which, in the case of Kyocera, was largely inspired by Dr. Inamori’s formative experience of having suffered a life-threatening illness, alongside the influence of Samurai Stoicism and Zen Buddhism, all of which moved him to relativize material values in favor of a life of purpose, similar to the teleological view elaborated in Plato’s Phaedo. -The second tier is substantiation — a set of rules for behavioral and financial management that Dr. Inamori considered crucial to achieving hypostasis. To this end, he designed the Amoeba Management System or AMS, in which the company is divided into “amoeba units,” each of which is responsible for its own profit and losses. In this way, the profit motive is passed down to the lowest management echelons. Referenced by Inamori management philosophy: putting faith in employees, IESE EDU https://lnkd.in/gxuCgbXT

1

The world of digital platforms keeps evolving as we define a 2.0 approach. Should governments accelerate their current roadmap towards digital self services?

4

We are looking forward to meet you at our event where we will share the future of accounting, current trends and challenges and how leading platforms such as #blackline, #trintech and #redwood provide answers on how to overcome those. #KPMG #intelligentfinance #automation

6