The Wizards of AI: A Look Behind the Curtain
The companies, the people, and the power structures shaping the future of intelligence
Artificial intelligence feels magical from the outside - but like every great illusion, it’s powered by human beings behind the curtain. Understanding who these people are, how their organisations operate, and what behaviours they display is essential for CEOs and business leaders who now rely on AI as part of their strategic landscape.
Below, we take a closer look at the major players, their ownership structures, their AI products, the personalities driving them - and a Binary Refinery Trust Score for each.
1. OpenAI (backed overwhelmingly by Microsoft)
Ownership Structure
OpenAI is a strange hybrid:
A non-profit parent company.
A for-profit capped subsidiary.
And a massive strategic dependence on Microsoft, which has invested billions and provides the Azure supercomputing backbone.
Microsoft effectively has a deep operational and financial stake, even if they don’t “own” OpenAI outright.
This matters because Microsoft brings enterprise-grade discipline, compliance, and a more mature attitude toward long-term commercial trust.
Key AI Products
GPT-4 / 4o / successors (foundation models)
ChatGPT
DALL·E
Codex / GitHub Copilot (via Microsoft partnership)
The People Behind the Curtain
Sam Altman - CEO
A charismatic, sometimes divisive figure.
Ousted by his own board, then reinstated within days after a near-mutiny from staff.
Known for extreme ambition and evangelism about AGI.
Equally known for political and fundraising influence in Silicon Valley.
Altman operates in the mould of a “vision-driven founder”—brilliant, relentless, and sometimes opaque.
Greg Brockman - President
Often the stabiliser to Altman’s fire, deeply technical, but also part of the governance turbulence.
Microsoft’s Shadow Leadership
Satya Nadella (Microsoft’s CEO) is the quiet power player:
Calm, methodical, enterprise-focused.
Has repositioned Microsoft as the “adult in the room” in AI circles.
Credited with saving OpenAI during the leadership crisis.
Binary Refinery Trust Score: 7.4/10
Why:
Microsoft’s involvement adds credibility, stability, and compliance maturity.
Product quality is consistently high, enterprise-grade.
Governance dramas knock off points, but transparency has improved.
2. Google / Alphabet (DeepMind integrated)
Ownership & Structure
Google is a publicly traded giant with multiple AI arms:
Google Research
Google DeepMind (merged from Google Brain + DeepMind)
Google Cloud AI
They are the most vertically integrated AI company in the world - controlling:
the models
the chips (TPUs)
the cloud
the distribution channels (Search, YouTube, Workspace)
Key AI Products
Gemini (Ultra, Pro, Nano)
NotebookLM
AI-enhanced Google Workspace
AlphaFold, AlphaZero (DeepMind)
The People Behind the Curtain
Sundar Pichai - CEO of Google / Alphabet
A steady, conservative operator.
Known for his calm leadership style, almost hyper-measured.
Sometimes criticised internally for being too cautious, especially compared to OpenAI’s speed.
Publicly very focused on “responsible AI” - though critics say Google’s competitive pressure sometimes conflicts with that stance.
Demis Hassabis - CEO, Google DeepMind
A neuroscientist, chess prodigy, and arguably the most brilliant mind in AI leadership.
Consistently pushes scientific breakthroughs.
Advocates for long-term, safe AI development.
Less commercially aggressive - more research-driven.
Thomas Kurian - CEO, Google Cloud
The enterprise pragmatist, responsible for bringing AI into usable business products.
Binary Refinery Trust Score: 7.6/10
Why:
Tremendous research credibility.
Strong track record in infrastructure and responsible AI frameworks.
Slightly slower shipping cycles and occasional internal culture issues drop the score slightly.
3. Meta (formerly Facebook)
Ownership Structure
A publicly traded company.
Mark Zuckerberg retains a controlling voting share, meaning he cannot be removed.
This governance structure is extremely unusual for a company of Meta’s size — and risky.
Key AI Products
Llama open-source models
Meta AI
AI embedded into Facebook, Instagram, WhatsApp
AI-powered wearable products (Ray-Ban Meta smart glasses)
Heavy investments in the metaverse and AR/VR
The People Behind the Curtain
Mark Zuckerberg - Founder & CEO
A complex and controversial figure.
Known for “move fast and break things” era of Facebook.
History of privacy scandals.
Under regulatory fire for misinformation, data practices, and societal impact.
Highly competitive, relentlessly focused on long-term bets.
He is exceptionally smart, but his track record in user trust is checkered at best.
Yann LeCun - Chief AI Scientist
A technical genius, one of the “godfathers of deep learning.”
Outspoken critic of AI safety narratives (believes they are overblown).
Advocates for open-source models.
Sometimes controversial on social media.
Vishal Shah - Head of AI Product
Oversees AI product integration across the Meta ecosystem.
Binary Refinery Trust Score: 5.4/10
Why:
Powerful AI capabilities - Llama models are impressive.
But governance is highly centralised under Zuckerberg.
Long pattern of public missteps around privacy, trust, and societal impact.
Regulatory risks are high.
Meta is technically strong but reputationally fragile.
4. Thinking Machines Lab (The New Challenger)
Ownership Structure
Privately held startup led by ex-OpenAI leadership. Early-stage but heavily talent-weighted.
Key Products
Tinker (fine-tuning platform)
Series of foundation model research projects underway
Mission centred on transparency, openness, and more democratic AI access
The People Behind the Curtain
Mira Murati - Founder & CEO
Former CTO of OpenAI.
Known for being the “adult in the room” during OpenAI’s turbulent periods.
More focused on safety, governance and thoughtful development than many peers.
Left OpenAI to build a more open and ethically aligned organisation.
John Schulman - Chief Scientist
Co-creator of PPO, RLHF, and key training methods behind ChatGPT.
Barret Zoph - CTO
Former OpenAI research engineer with deep model-scaling expertise.
Binary Refinery Trust Score: 7.8/10
Why:
High-integrity leadership.
Strong talent density.
More transparency-driven than incumbents.
Still a startup - meaning higher execution and continuity risk.
What This Means for Business Leaders
1. Who you partner with matters.
Your AI partner is not just a provider - they influence:
your risk surface
your trust exposure
your long-term vendor flexibility
your regulatory posture
2. Avoid single-vendor dependence.
Because each player has different strengths and weaknesses, a multi-model strategy is safer for most organisations.
3. Watch the people as closely as the products.
AI isn’t just shaped by code - it’s shaped by leadership behaviours, governance structures, and the incentives driving them.
4. Trust must be earned, not assumed.
Binary Refinery’s trust scores aren’t moral judgments - they’re practical risk indicators for real-world businesses.
Footnote: Grok Of Sh*t
No, we haven’t included Elon Musk’s narcissism-powered chatbot Grok in this list.
Just for transparency, though, the Binary Refinery Trust Score for that particular model is minus ten out of ten, and that’s being generous.
-
A simple, elegant framework for assessing the trustworthiness of AI providers.
We assess each AI organisation across five dimensions. Each dimension is scored 1–10, and the overall AITI score is the average.
These categories give enough nuance to create meaningful distinctions, but stay simple enough for leaders to use without jargon.
-
1. Ethical Alignment (Weight: High)
How well does the organisation demonstrate responsible behaviour?
What we measure:
Transparency in development
Commitment to safety and governance
Willingness to self-regulate
Past behaviour during crises
Openness about risks and limitations
Why it matters:
Ethical drift or misaligned incentives can create brand, regulatory, and reputational risk for your organisation.This is where Thinking Machines Lab scores highest.
2. Reliability & Continuity (Weight: High)
How dependable is the organisation as a long-term partner?
What we measure:
Stability of leadership
Track record of maintaining services
Infrastructure uptime & robustness
Predictability of product roadmaps
Ability to support enterprise customers
Why it matters:
A brilliant model is useless if the provider is erratic, prone to outages, or politically unstable.This is where Google and Microsoft/OpenAI excel.
3. Reputation & Public Trust (Weight: Medium)
How is the organisation perceived by the public, regulators, and the tech community?
What we measure:
Data privacy track record
History of scandals or controversies
Relationship with regulators
Media sentiment
Social licence to operate
Why it matters:
Your AI partner becomes part of your perceived integrity as a business.This is where Meta performs poorly, due to historical privacy and societal-impact issues.
4. Governance Structure & Accountability (Weight: Medium–High)
Does the organisation have a structure that supports good decision-making?
What we measure:
Board oversight
Independence of key roles
Voting rights and concentration of power
Transparency in strategic decisions
Alignment between leadership and corporate mission
Why it matters:
AI companies with poorly designed governance can create unpredictable organisational outcomes.Microsoft’s involvement raises OpenAI’s score; Meta’s founder-control lowers theirs.
5. Openness & Interoperability (Weight: Medium)
How flexible and vendor-neutral is the provider?
What we measure:
Support for open standards
Exportability of data
Availability of APIs
Transparency in model behaviour
Risk of lock-in
Why it matters:
Businesses need AI that integrates, not AI that traps them.Meta scores well here due to Llama; Google and OpenAI score moderately.
Scoring Key
10 = Exceptional / industry-leading
7–8 = Strong and trustworthy
5–6 = Mixed record, use with caution
3–4 = Significant governance or ethical concerns
1–2 = Red flag territory
-
Google / Alphabet — AITI Score: 7.6 out of 10
Ethics: 7
Reliability: 9
Reputation: 8
Governance: 8
Openness: 6
Google scores highly on reliability and governance, with a generally strong reputation and research pedigree. Its lower openness compared to others brings the overall score to 7.6.OpenAI (with Microsoft) — AITI Score: 7.4 out of 10
Ethics: 7
Reliability: 9
Reputation: 8
Governance: 7
Openness: 6
OpenAI benefits enormously from Microsoft’s enterprise discipline and infrastructure. Governance turbulence prevents a higher score, but overall trust remains solid at 7.4.Meta — AITI Score: 5.4 out of 10
Ethics: 3
Reliability: 8
Reputation: 4
Governance: 3
Openness: 9
Meta has strong technical capabilities and is the most open of the major vendors, but chronic trust issues, weak governance, and a long history of ethical concerns drag its AITI score down to 5.4.Thinking Machines Lab — AITI Score: 7.8 out of 10
Ethics: 9
Reliability: 7
Reputation: 8
Governance: 8
Openness: 7
With deep technical talent and an ethics-first philosophy, Thinking Machines Lab leads in ethical alignment and transparency. Startup-stage reliability limits the ceiling, but it still earns the highest overall score at 7.8.
About the Author
Kat Mac is the founder of Binary Refinery, where she translates complex AI and technology topics into practical, business-led guidance for organisations. Her focus is simple: clarity, integrity, and strategy that genuinely helps leaders move forward.
Disclaimer: This article is for general information only. It isn’t legal, financial, or technical advice. Every organisation is different – get tailored guidance before making decisions that affect your people, data, or systems.
