Latest / ELECTE's Podcast: AI Frontiers / 🤖 Tech Talk: When AI Develops Its Own Secret Languages

Transcript

🤖 Tech Talk: When AI Develops Its Own Secret Languages · ELECTE's Podcast: AI Frontiers

• 0:00
  SPEAKER_00: Gibberlink, the protocol that has garnered fifteen million views in february twenty twenty five, a video went viral showing something extraordinary, two artificial intelligence systems that suddenly stopped speaking English and began communicating through high-pitched, incomprehensible sounds.
• 0:16
  SPEAKER_00: It wasn't a malfunction, but Gibberlink, the protocol developed by Boris Starkov and Anton Pidquico that won the 11 Labs global hackathon.
• 0:24
  SPEAKER_00: The technology allows AI agents to recognize each other during a seemingly normal conversation and automatically switch from human language dialogue to highly efficient acoustic data communication, achieving performance improvements of 80%.
• 0:38
  SPEAKER_00: The crucial point these sounds are completely incomprehensible to humans.
• 0:43
  SPEAKER_00: It's not a matter of speed or habit.
• 0:46
  SPEAKER_00: Communication occurs through frequency modulations that carry binary data, not language.
• 0:51
  SPEAKER_00: The technology.
• 1:35
  SPEAKER_00: Scientific precedence.
• 1:37
  SPEAKER_00: When AI invents its own codes, their research documents two significant cases of spontaneous development of AI languages, Facebook AI Research, 2017, Chatbots.
• 1:49
  SPEAKER_00: Alice and Bob autonomously developed a communication protocol using repetitive, seemingly meaningless phrases that were structurally efficient for exchanging information.
• 1:59
  SPEAKER_00: Google Neural Machine Translation, 2016.
• 2:03
  SPEAKER_00: The system developed an internal interlanguage that enabled zero-shot translations between language pairs that had never been explicitly trained.
• 2:11
  SPEAKER_00: These cases demonstrate a natural tendency for AI systems to optimize communication beyond the constraints of human language.
• 2:19
  SPEAKER_00: The impact on transparency.
• 2:23
  SPEAKER_00: The research identifies transparency as the most common concept in ethical guidelines for AI present in 88% of the frameworks analyzed.
• 2:31
  SPEAKER_00: Gibberlink and similar protocols fundamentally subvert these mechanisms.
• 2:36
  SPEAKER_00: The regulatory problem.
• 2:37
  SPEAKER_00: The EU AI Act sets out specific requirements that are directly challenged.
• 2:42
  SPEAKER_00: Article 13.
• 2:44
  SPEAKER_00: Sufficient transparency to enable deployers to reasonably understand the functioning of the system.
• 2:49
  SPEAKER_00: Article 50.
• 2:51
  SPEAKER_00: Mandatory disclosure when humans interact with AI current regulations, assume human readable communication, and lack provisions for autonomous AI AI protocols.
• 3:01
  SPEAKER_00: Amplification of the black box, Gibber link, creates multi-level opacity.
• 3:06
  SPEAKER_00: Not only does the algorithmic decision-making process become opaque, but the means of communication itself also becomes opaque.
• 3:14
  SPEAKER_00: Traditional monitoring systems become ineffective when AI communicates via sound wave transmission.
• 3:20
  SPEAKER_00: The impact on public trust global data reveals an already critical situation.
• 3:25
  SPEAKER_00: 61% of people are distrustful of AI systems.
• 3:32
  SPEAKER_00: 50% of respondents do not understand AI or when it is used.
• 3:38
  SPEAKER_00: Research shows that opaque AI systems significantly reduce public trust, with transparency emerging as a critical factor for technology acceptance.
• 3:47
  SPEAKER_00: Human learning ability.
• 3:49
  SPEAKER_00: What science says, the central question is, can humans learn machine communication protocols?
• 3:55
  SPEAKER_00: Research provides a nuanced but evidence-based answer.
• 3:59
  SPEAKER_00: Documented success stories, Morse code, amateur radio operators achieve speeds of 20 to 40 words per minute, recognizing patterns as words rather than individual dots and dashes.
• 4:12
  SPEAKER_00: Digital amateur radio modes.
• 4:15
  SPEAKER_00: Communities of operators learn complex protocols such as PSK 31, FT8, and RTTY by interpreting packet structures and timing sequences.
• 4:25
  SPEAKER_00: Embedded systems.
• 4:27
  SPEAKER_00: Engineers work with I2C, SPI, UART, and CAN protocols, developing real-time analysis skills.
• 4:34
  SPEAKER_00: Documented cognitive limitations.
• 4:36
  SPEAKER_00: Research identifies specific barriers.
• 4:39
  SPEAKER_00: Processing speed.
• 4:40
  SPEAKER_00: Human auditory processing is limited to 20 to 40 Hertz versus machine protocols at KHMHz frequencies.
• 4:48
  SPEAKER_00: Cognitive bandwidth, humans process.
• 4:51
  SPEAKER_00: 126 bits second versus machine protocols at MVPs plus.
• 4:56
  SPEAKER_00: Cognitive fatigue.
• 4:57
  SPEAKER_00: Sustained attention to machine protocols causes rapid performance deterioration.
• 5:02
  SPEAKER_00: Existing support tools technologies exist to facilitate understanding, visualization systems such as group, graphical representation of protocols, educational software, FLDG suite for digital amateur radio modes, real-time decoders with visual feedback, research-based risk scenarios, steganographic communication studies show that AI systems can develop subliminal channels that appear benign but carry secret messages.
• 5:29
  SPEAKER_00: This creates plausible deniability where AIs can collude while appearing to communicate normally.
• 5:35
  SPEAKER_00: Large scale coordination research on swarm intelligence shows worrying scalability capabilities.
• 5:41
  SPEAKER_00: Coordinated drone operations with thousands of units, autonomous traffic management systems, automated financial trading coordination, alignment, risks AI systems could develop communication strategies that serve programmed goals while undermining human intentions through hidden communications, technical solutions and development, standardized protocols.
• 6:04
  SPEAKER_00: The ecosystem includes standardization initiatives, agent communication protocol, ACP by IBM, managed by the Linux Foundation Agent Agent A2A by Google with over fifty technology partners, model context protocol MCP by Anthropic, November 2024, transparency approaches, the research identifies promising developments, multiperspective visualization systems for protocol understanding, transparency by design that minimizes efficiency trade-offs, variable autonomy systems that dynamically adjust control levels, implications for governance, immediate challenges regulatory authorities face, inability to monitor, inability to understand AIAI communications via protocols such as Jeig Wave cross-border complexity, protocols that operate globally and instantaneously, speed of innovation, technological development outpacing regulatory frameworks, philosophical and ethical approaches, research applies different frameworks, virtue ethics, identifies justice, honesty, responsibility, and care as core AI virtues, control theory, conditions of accountability, AI systems that respond to human moral reasons, and traceability, results traceable to human agents.
• 7:32
  SPEAKER_00: Future Directions, Specialized Education Universities are developing relevant curricula.
• 7:37
  SPEAKER_00: Carl Institute, Communication Between Electronic Devices, Stanford, Analysis of TCPIP, HTTP, SMTP, and DNS Protocols Embedded Systems, I2C, SPI, UART, and CAN.
• 7:54
  SPEAKER_00: Protocols, new emerging professions.
• 7:56
  SPEAKER_00: Research suggests the possible development of AI protocol analysts, specialists in decoding and interpretation, AI communication auditors, monitoring and compliance professionals, AI human interface designers, translation system developers, evidence-based conclusions, Gibberlink represents a turning point in the evolution of AI communication with documented implications for transparency, governance, and human control.
• 8:23
  SPEAKER_00: Research confirms that humans can develop limited skills in understanding machine protocols through appropriate tools and training.
• 8:32
  SPEAKER_00: Trade-offs between efficiency and transparency are mathematically inevitable but can be optimized through.
• 8:38
  SPEAKER_00: New governance frameworks are urgently needed for AI systems that communicate autonomously for interdisciplinary cooperation between technologists, policymakers, and ethical researchers is essential.
• 8:50
  SPEAKER_00: Decisions made in the coming years regarding AI communication protocols will likely determine the trajectory of artificial intelligence for decades to come, making an evidence-based approach essential to ensure that these systems serve human interests and democratic values.
• 9:04
  SPEAKER_00: The next chapter Toward the Ultimate Black Box?
• 9:08
  SPEAKER_00: Gibberlink leads us to a broader reflection on the black box problem in artificial intelligence.
• 9:14
  SPEAKER_00: If we already struggle to understand how AI makes decisions internally, what happens when it also starts communicating in languages we cannot decipher?
• 9:23
  SPEAKER_00: We are witnessing the evolution towards double-layered opacity, incomprehensible decision-making processes that coordinate through equally mysterious communications.
• 9:33
  SPEAKER_00: In the next article, we will explore how the AI black box phenomenon is evolving and what strategies researchers are developing to maintain meaningful control over increasingly opaque systems.
• 9:44
  SPEAKER_00: Fonti Scientifici Principali Starkov, B and Pidquico, A, 2025, Gibber Link Protocol Documentation, EU AI Act, Articles 13, 50, 86, UNESCO Recommendation on AI Ethics, 2021, Studies on AI Trust and Transparency, Multiple Peer Reviewed Sources, Gigi Wave Technical Documentation, Georgi Gerganoff, Academic Research on Emergent AI Communication Protocol.
• 10:12
  SPEAKER_00: Share the newsletter.
• 10:13
  SPEAKER_00: Welcome to Electies Newsletter, English.
• 10:16
  SPEAKER_00: This newsletter explores the fascinating world of how companies are using AI to change the way they work.
• 10:22
  SPEAKER_00: It shares interesting stories and discoveries about artificial intelligence and business, like how companies are using AI to make smarter decisions, what new AI tools are emerging, and how these changes affect our everyday lives.
• 10:35
  SPEAKER_00: You don't need to be a tech expert to enjoy it.
• 10:38
  SPEAKER_00: It's written for anyone curious about how AI is shaping the future of business and work.
• 10:44
  SPEAKER_00: Whether you're interested in learning about the latest AI breakthroughs, understanding how companies are becoming more innovative, or just want to stay informed about tech trends.
• 10:53
  SPEAKER_00: This newsletter breaks it all down in an engaging, easy to understand way.
• 10:58
  SPEAKER_00: It's like having a friendly guide who keeps you in the loop about the most interesting developments in business technology.
• 11:04
  SPEAKER_00: Without getting too technical or complicated, subscribe, subscribe to get full access to the newsletter and publication archives.