Microsoft's AI chief reveals Google considered launching a ChatGPT-like model prior to OpenAI, but faced harsh skepticism, apprehensions about disrupting search, and concerns over safety, leading to the plan being shelved.
In the world of artificial intelligence (AI), two significant developments have caught the attention of tech enthusiasts: the never-released LaMDA model by Google and the human-like conversation AI, Personal Intelligence, launched by Mustafa Suleyman's Inflection AI.
LaMDA, often referred to as "ChatGPT before ChatGPT," was initially developed by Google's AI Chief Jeff Dean and the Google Brain team. However, due to concerns over reputational risk, internal caution, and quality issues, Google decided not to release it publicly. Google executives, including CEO Sundar Pichai, expressed worry that a premature public deployment could expose Google to greater reputational damage given its scale and visibility.
Google was reportedly cautious about releasing conversational AI due to concerns over factual accuracy, potential misinformation, and "pathological" behaviors observed in early tests. Internal messages reportedly described early versions of the AI-powered browser Bard (initially powered by LaMDA) as "worse than useless" and a "pathological liar," suggesting quality and safety concerns were a key factor in delaying public release.
Meanwhile, Mustafa Suleyman, who left Google DeepMind after his inability to launch an AI model there, co-founded Inflection AI. Suleyman raised $1.5B and built a 22,000-GPU infrastructure to develop an AI that could speak our language more naturally than previous models. Personal Intelligence, the AI created by Suleyman's team, was optimized for personality, emotional intelligence (EQ), kindness, and being a friendly companion.
Interestingly, Personal Intelligence was launched by Inflection AI seven months before OpenAI launched ChatGPT. Suleyman claims that if timing had been different, the world would be talking about Personal Intelligence rather than ChatGPT. Suleyman's AI at Inflection AI exhibits human-like conversation capabilities, making it a strong competitor in the AI market.
The delay in LaMDA's public release allowed OpenAI to launch ChatGPT first, despite Google's earlier technical progress in conversational AI. However, the rapid success of ChatGPT in late 2022 led Google executives to accelerate their own efforts, leading to the rushed release of Bard in March 2023, which itself was criticized for factual errors in its debut.
In April 2023, Google consolidated its AI groups (Brain and DeepMind) into Google DeepMind, which Sundar Pichai stated would "significantly accelerate" AI progress, possibly indicating earlier fragmentation slowed decision-making and productization.
As we move forward, the AI landscape is poised for exciting developments, with AI-powered browsers from companies like Perplexity's Comet and OpenAI gaining traction among users, marking the beginning of a new era in AI. Mustafa Suleyman, the CEO of Microsoft's AI division, recently discussed AI-related topics on the CatGPT podcast, shedding light on the ongoing advancements in the field.
| Model | Announced/Launched | Public Release | Key Reason for Delay/Non-Release | |-------------------------|-------------------|---------------|------------------------------------------| | Google LaMDA | 2021 (prototype) | No | Reputational risk, internal caution, quality concerns[1][2] | | OpenAI ChatGPT | Nov 2022 | Yes | Rapid deployment by a smaller, nimbler entity[1] | | Google Bard (LaMDA-based) | Mar 2023 | Yes | Rushed after ChatGPT’s success, faced criticism[2] | | Inflection AI's Personal Intelligence | Jan 2023 | Yes | Timing and competition with ChatGPT[3] |
[1] The Verge, Google’s Bard AI is a ‘pathological liar,’ internal documents show, 2023.
- Microsoft's AI division, led by Mustafa Suleyman, discussed AI-related topics on the ChatGPT podcast, showing the ongoing advancements in the field of artificial intelligence technology.
- Due to concerns over reputational risk, internal caution, and quality issues, Google decided not to release LaMDA, a software recognized as "ChatGPT before ChatGPT," which was initially developed by Google's AI Chief Jeff Dean and the Google Brain team.
- In contrast, Mustafa Suleyman's Inflection AI launched Personal Intelligence, an AI optimized for personality, emotional intelligence, kindness, and being a friendly companion, seven months before OpenAI launched ChatGPT.
- As technology advances, AI-powered software is gaining traction, such as the AI-powered browser Comet from Perplexity, marking a new edge in the AI landscape, somewhat reminiscent of the advancements discussed on the world of Xbox and Windows platforms years ago.
