mAIndset

• What AI PRACTITIONERS Comment Lately ON 2025 TRENDS

  Over the last 6 weeks, Filip has travelled the Globe to visit 5 major AI conferences. He also collected key points that have been discussed. And thus we told ourselves, it would be a great opportunity to maybe summarize also for you, our audience, what the AI Practitioners are saying regarding the 2025 development in the AI domain. Therefore, in this 15th episode of the mAIndset podcast, we gradually cover:01:40 Ai Experts: We Don't Need Bigger And Bigger AI Models03:29 Shall We Rather Lean On SLM's ?!06:40 Taking 8-Wheel Truck For Grocery Shopping07:50 OpenAI Still Plans Overkill Of New Model Releases11:03 Stop Thinking About AI Capabilities !13:20 2025 Dictionary Word of Year: "(AI) Agents"15:17 Some Still On Wrong Track: Metering AI Agents By Numbers16:40 CONFIRMED: Agents Take Mainly Junior Jobs18:25 Efficiency Lever At Whooping 1:67 Rate20.36 Expect Top Managers To Have AI Agents Goals for 202623:04 Soon There Will Be Explosion In Token Consumption We Aren't Ready For25:02 Additional Nuclear Powerplant for Slovak AI27:24 New Local-PC-running-LLM Chips (from NVIDIA)31:56 Green Considerations In AI: 2 PDFs in GPT Equals ...35:42 So What Shall We Do Differently With AI ?!36:47 Do We Step Back From AI Search To Google Search ?38:03 Caching LLM Can Save 40% Of Your Costs39:16 Uff, Now We Will Store Tokens Into Traditional Databases?41:02 ALREADY MOST USED AI Use Cases Are ...44:17 SW Developer Roles Now Redefines To Test-Shaping46:26 OpenAI Push-back "Resistance" Is GatheringIf you want to suggest (or vote up) topics for next episodes of the mAIndset podcast, go and vote here:https://maindsetpodcast.com/If you have a question or comment, don't hesitate to reach out on:[email protected]

  --------  

  51:04

  --------

  51:04
• AI BUBBLE - Hints How To Track If It's Coming

  Topic of AI Bubble started to surface already last year and comes back in (ever increasing) frequent hops. Studies like infamous MIT study of 95% AI Failing are loading the discussion even further. So we settled down with David for discussion on where the truth might be on this Bubble issue. To avoid being just shallow, we decomposed the total Revenue (and Profit) profiles of companies and discussed in deep detail the development around AI companies' User Bases, Intensity of Use and Frontier Labs' Pricing power. The development in the very topic is so dynamic that next events my render these kind of talks soon irrelevant. Therefore, we invite you watch through quickly :) and to extract the measuring logic rather than just one snapshot conclusion. Enjoy!Chapters:01:40 Do We Have, AI Bubble Or Just Mind Games ?04:10 M7 Taking Lion Share of Stock Market06:25 It's Complicated, Because08:40 AI Is Surely Getting To Big For Traditional Investors10:10 Most Investors Are Locked Out Of Investment In AI12:34 Reading The Recent Quarterly Results, We See ...15:40 Profits, Not Reported, But Once Can Assume19:54 When The Tide Goes Down, It Is Clear Whole Had Swimsuit And Who Didn't21:02 Introducing 3 Factors: User Base + Intensity + Pricing25:22 Positive And Negative Factors Of User Base Dynamics28:07 Intensity of Use Is Tricky to Estimate31:40 Wow Effects Try To Push Usage, But Stumbling On This35:03 Free Tier Is Still Bloating Impression Of Demand38:04 Quite Bluntly : AI Is Still Too Cheap41:00 30USD+ Would Be Psychological Limit For Me44:04 Bring Your Own AI To Work Might Play A Role Here47:51 After Here All The Facts David Changes His Opinion

  --------  

  50:31

  --------

  50:31
• Where AI Meets Medicine On Crossroads [with Prof. Amina Qutub]

  Having distinguished Prof. Amina A. QUTUB for interview is a dream of many Tech podcasters (or it certainly should be). As only few people have both the vision to see where AI is heading and the hands-on expertise to push those breakthroughs into the real life of clinics and hospitals. Her work is on intersection of high-end, expert Biology and Computational Systems and even both ways: With her team she not only builds models that replicate human brain BUT ALSO uses these insights to gear "regular" AI (as we all know it from LLMs) to save lives in Emergency Rooms and help Clinicians to stay On-Top-Of-The-Game even in long shifts. Thus, talking to someone, who's inventions can literally save your own life in future made this discussion so eye-opening and awesome at the same time. With David, we definitely took a chance to address the AI + Medicine combo from several angles and got from Dr. Qutub handful of insights nuggets. So, go ahead & explore all chapters of this great interview:01:50 Patterns of Communication, From Molecules to Humans02:40 Where To Look For AI Start-Up Inspiration In Medicine04:55 Turing Test in Medical AI Would Need to Include Smell and Taste06:56 Combining Many Clinicians On Top Of Their Game09:15 Accessibility Paves The AI Adoption (and hence hinders Medical AI)11:24 In Full Emergency Room AI Split-second Decision Really Helps16:35 Originally Deep Learning Inspired By Brain, But Where Do We Go Now20:35 Is There Hope For Humanity To Merge Bio Brain Efficiency With AI26:08 Building AI That Does Not Hurt Or Kill29:34 How Do AI Scientists Choose (from ever changing) Top AI Models32:05 Where Can/Should We Augment Humans With AI (and where to stop in it)35:50 Launching AI Is Not Like Introducing Penicillin38:19 Humble Self-Regulation (In Times Where Regulators Not Fully Grasp AI Principles)41:30 AI Is Useful Even In My Private Life, But I Miss ...44:16 What If We Invest Into Creating Super-Humans ?Amina Ann Qutub, PhD is the Burzik Professor of Engineering Design and Associate Professor of Biomedical Engineering at the University of Texas, San Antonio. She serves as the Assistant Director of Strategic Alliances for the MATRIX AI Consortium and a research thrust lead for the Augmenting Human Performance thrust. Dr. Qutub is also the Director of the UTSA – UT Health San Antonio Graduate Group in Biomedical Engineering and co-lead of the Center for Precision Medicine. Dr. Qutub is pioneering methods at the interface of computer science, biology and engineering to study the design of human cells, and help eradicate diseases affecting cells of the brain and vasculature. In new translational work, Dr. Qutub is co-lead (with Drs. Brian Eastridge, MD, UTHSCSA and Alan Cook, MD, UTTyler) of the iRemedyACT project to develop AI tools that can minimize time to treatment and optimize care for trauma patients.

  --------  

  47:05

  --------

  47:05
• AI Browsers | Where Do Work Well And What Can't They Do Yet

  After years of consuming the AI mainly through chatbots and custom applications, finally we AI makes itself available also in form of the Web/AI Browsers. But that is where the story only starts, as it opens plethora of new questions:Is AI browser a completely new way to consume public content or are we longing just for the AI plug-ins into regular browsers. How will AI environment deal with things like Ad-blockers and privacy controlShall we crave for Ai browsers to be all-mighty agents doing thins for us or just from side watching advisor without any agencySpeaking of AI agents, will they swallow this early Ai browser attemptsShall one count the AI Browsers to become separate industry and which players are trying to dominate this space. All those topics and more you can hear from the recent episode of the www.maindsetpodcast.com

  --------  

  35:36

  --------

  35:36
• 2025 AI Turns Nobody Saw Coming (but they still happened)

  We started 2025 year with mAIndset podcast episode "2025 AI Surprises You Are Not Ready For" outlining prediction about where AI will move in next 12 months. As we are after H1 results of all key players, let's look back on what predictions are turning true and where trends go differently as originally anticipated.

  --------  

  39:50

  --------

  39:50

Weitere Technologie Podcasts

Trending Technologie Podcasts

Über mAIndset