Julio Serrano

🇸🇻 in 🇫🇮
Advanced Technologies for Business Strategy Doctoral Researcher | B2B Marketing specialist | AI & ML 4 Business | ⛵️ Baltic Sailor ⚙️ex-Rotaractor

[Julio Serrano @julio_serrano89]

Is siloed data killing your AI ROI? 83% of CDOs trust AI agents, yet only 26% can power new revenue with their current data. Strategy is shifting from "collecting all" to "deploying on a mission". #AI #DataStrategy

PDF: https://t.co/UGA8O4ZlJE

[Julio Serrano @julio_serrano89]

Is deep learning an illusion? Google’s new Nested Learning (NL) paradigm says yes. By mimicking human brain wave frequencies, their HOPE model outperforms Transformers in reasoning. It’s AI that learns like we do.
#AI #GoogleResearch

PDF: https://t.co/libfv1kRS9

[Julio Serrano @julio_serrano89]

AI reached 1.2B users in <3 years, beating the smartphone & internet. But a new divide is here: adoption in the Global North is 2x the Global South. 4B people still lack the basic electricity or internet to even try it. #AI #TechTrends

PDF: https://t.co/ypy0cGe5Uh

[Julio Serrano @julio_serrano89]

AI isn't replacing strategic foresight; it is becoming the ultimate sparring partner.

New OECD data shows 69% of experts now use AI for trend analysis to stress-test human ideas rather than replace them.

#StrategicForesight #AIGovernance

PDF: https://t.co/10mmy4dNYA

[Julio Serrano @julio_serrano89]

Human vs. AI performance.

Stop using AI for peer review judgment. 🛑

The data is clear: while 84% of researchers use AI, the consensus is that humans VASTLY outperform bots in review quality.

#PeerReview #Science

PDF: https://t.co/4sIHuxukyj

[Julio Serrano @julio_serrano89]

This EU law could save businesses €1B/year by killing off redundant red tape. Major move: repealing the P2B Regulation & consolidating data rules into just 2 acts. Say goodbye to cookie banner fatigue with automated privacy choices.

#EU #DigitalAct

PDF: https://t.co/iJNk1awSbW

[Julio Serrano @julio_serrano89]

Is AI truly reasoning, or just memorizing the internet?

Humanity's Last Exam (HLE) is the new gold standard for AI. 2,500+ graduate-level questions that frontier models like GPT-4o and o1 still struggle to pass.

#AI #LLM #HLE

PDF: https://t.co/i1qGwH51Zq

[Julio Serrano @julio_serrano89]

This study shows workers slash AI use by 14% when a manager is watching. We fear it signals a "lack of confidence" in our judgment. We are choosing lower performance just to protect our image.

#AI #WorkplaceTrends

PDF: https://t.co/QUw9egEJsH

[Julio Serrano @julio_serrano89]

We are governing XXI century AI with XX century laws. (It’s a recipe for disaster) Experts warn the window to prevent irreversible catastrophe is closing fast. We don't need more "ethics principles." We need an IAEA for algorithms

#AIGovernance #AI

PDF: https://t.co/nF1lgCHFSX

[Julio Serrano @julio_serrano89]

AI just beat the human world record for the "Packing Circles" problem. ⭕️

AlphaResearch discovered a new algorithm (score: 2.939) by simulating peer reviews and executing code loops.

It is discovering the unknown.

#AlphaResearch #AI

PDF: https://t.co/YP0ph4xa6E

[Julio Serrano @julio_serrano89]

Can AI agents discover what humans missed?

The STATION is an open-world where AI agents autonomously publish papers; code; and collaborate. It just hit SOTA in 5 domains (beating AlphaEvolve and LLM-TS).

#AI #Innovation

PDF: https://t.co/YXlOAVnO1A

[Julio Serrano @julio_serrano89]

Can AI learn like a student? MIT’s new "SEAL" method lets LLMs write their own study sheets to permanently memorize new info. Accuracy jumped 15% and skill-learning soared by 50%. The era of static AI is over.
#AI #MIT

Article: https://t.co/axguGsM3lV

[Julio Serrano @julio_serrano89]

Can Finland lead Europe's AI transition? The SW4E AI Sandbox enables 600x faster experimentation while staying 100% EU AI Act compliant. It offers 25+ services for secure, transparent, and collaborative AI development. #AI #Finland

PDF: https://t.co/NS2P3X8KQH

[Julio Serrano @julio_serrano89]

Is "powerful AI" the only real threat? CSET research shows AI harm correlates weakly with model capabilities. Many incidents involve simple, single-purpose systems. We need diverse safety tools, not just technical patches.

PDF: https://t.co/OlELUtJbqx

[Julio Serrano @julio_serrano89]

94% of CEOs think AI could give better advice than their human board members. A recent study shows AI agents outperform humans in decision quality and structural discipline. Is the era of the human-only board officially over? #AI #TechTrends

Source: https://t.co/ZbG3wmxKFR

[Julio Serrano @julio_serrano89]

31% of small firms now use AI; with 65% seeing performance gains. Surprisingly; 83% report NO change in staff needs (it is a tool; not a replacement). Is your business part of the 31%? #SMEs #GenerativeAI

PDF: https://t.co/npvYklL8RR

[Julio Serrano @julio_serrano89]

Can AI actually "think" before it acts? CrewAI is proving it can by reaching 95% accuracy through "System 2" reasoning. Instead of one bot: it uses a team of agents to solve ethical dilemmas and complex tasks.

#AI #CrewAI

PDF: https://t.co/zThq3qbfEW

[Julio Serrano @julio_serrano89]

Can 100 lines of code do the work of 430?

Microsoft’s AutoGen allows multiple AI agents to "talk" to solve tasks. It hit 69% accuracy on the MATH dataset: crushing vanilla GPT-4’s 55%.

It’s not just a bot; it’s a workforce.

#AI #AutoGen

PDF: https://t.co/I5M9lolLyJ

[Julio Serrano @julio_serrano89]

This research shows agentic AI is now autonomously designing experiments and analyzing data in chemistry and biology. While literature reviews still challenge them, the speed of discovery is exploding.

#AI #Science

PDF: https://t.co/1lYbXtCuFi

[Julio Serrano @julio_serrano89]

Is your B2B strategy ready for the $3.9T AI revolution? Research shows AI is moving from "human support" to making independent, effective decisions. To lead in 2026, you must pivot to digital intelligence models now.

#B2B #AI

PDF: https://t.co/653e2W7OZp