The arms race between the two tech giants to develop and deploy increasingly sophisticated AI models will likely have a big impact on the marketing industry – and on society at large
Which company will rule the future of search?
This is a question that’s come increasingly to the fore since generative artificial intelligence (AI) models like ChatGPT, Dall-E 2 and Midjourney have erupted into the zeitgeist in recent months. Their sudden and meteoric rise (it only took ChatGPT around two months to reach 100m monthly active users) has upended the widely held belief that advanced AI is a purely sci-fi concept. Today, the world is waking up to the fact that this technology has in fact arrived – and it’s advancing rapidly, with potentially huge implications for industries as far afield as medicine, law, journalism and marketing.
Many brands, naturally, have been paying close attention to the rise of generative AI. Some, like Duolingo and Expedia, have launched their own chatbots that are powered by GPT-4, the large language model (LLM) behind ChatGPT. Others, like Coca-Cola, have embraced image-generating AI models to expand their brand’s artistic aesthetic and foster a sense of collaboration and co-creation between themselves and independent artists. And some leading tech companies, like Meta and Amazon, have recently announced that they’ve begun their own efforts to innovate with generative AI.
But the business sector that has perhaps undergone the most significant changes since generative AI has gone mainstream is internet search.
That sector’s longtime leaders, Google and Microsoft, have for years leveraged AI as a core component of their respective search engines; it’s algorithms, not human beings, that determine which search results are prioritized over others. The advent of generative AI, however, arguably marks a phase transition for the search industry. In GPT-4, many within these companies see the possibility of creating a more engaging, personalized and robust online search experience.
Anna Bofa, a former Google and Facebook employee who founded her own company, Crate, in 2021, says that Google and Microsoft are “not so much creating search engines as they are creating ‘call-and-response’ engines.” In other words: Systems that are based on reciprocal dialogue between humans and AI, rather than one-way systems which algorithmically spit out a menu of search results.
A dangerous road ahead?
There are some who worry that the corporate race to develop and deploy increasingly sophisticated AI models is a dangerous process that could potentially spiral out of control and lead to civilizational collapse.
In March, a cohort of tech luminaries, AI researchers and experts hailing from other fields joined forces to publish an open letter that made the case that our current trajectory is leading to “ever more powerful digital minds that no one – not even their creators – can understand, predict, or reliably control.” The letter, published by the Future of Life Institute, argued that all AI labs should halt “the training of AI systems more powerful than GPT-4” for a minimum of six months in order to give the industry and regulators an opportunity to implement effective guardrails and chart a path forward that will not lead to disaster. (Whether or not six months would really be enough time to achieve such goals is a matter of debate.)
“A lot of people have said for many years that there will come a time when we want to pause [AI research] a little bit,” MIT physicist Max Tegmark, one of the signatories of the open letter, recently told the podcaster Lex Fridman. “That time is now.”
While there are undoubtedly some within Google and Microsoft who share Tegmark’s concerns – that AI research is outpacing our ability to implement safeguards – the dynamics of capitalism have thus far proven to be more powerful. As Tegmark points out in his most recent appearance on Fridman’s podcast, it’s highly unlikely that one major tech company that’s developing AI is going to stop if its competitors are continuing to move full steam ahead; such a move probably wouldn’t sit well with most shareholders, who tend to be primarily concerned with revenue.
The only viable option, Tegmark believes, is a pause that’s unilaterally agreed upon by the major tech companies that are building new AI models.
A new era
As a concept, AI has been around since the 1950s, and It’s now been more than 25 years since IBM’s Deep Blue – an AI model – beat chess grandmaster Garry Kasparov. Today, smartphones and digital assistants like Amazon’s Alexa have made AI an almost essential part of many of our lives; it curates our social media feeds, shows us the driving routes with the least amount of traffic when we’re using Google Maps, and even speaks to us in strange, disembodied female voices.
But the past few months have arguably marked a fundamental shift in the relationship between humans and AI; there’s a sense that the ground is shifting beneath our feet. Artificial general intelligence (AGI) – computer systems that are at least as intelligent as human beings across a wide range of tasks – may not be here yet, but there are some who believe it could be just over the horizon. Indeed, a research paper published by Microsoft in March, in which researchers from the company ran experiments with GPT-4 to test the model’s capabilities, found “that it could reasonably be viewed as an early (yet still incomplete) version of an [AGI] system.”
Future generations could very well look back on this historical moment and recognize it as the dawning of a new era.
To give you a sense of just how rapidly this shift is occurring, here’s a brief timeline of some of the most notable recent AI-related developments from Google and Microsoft (note: this is by no means intended to be an exhaustive list of all of the major recent breakthroughs in the AI industry at large – there are many smaller companies that are making significant contributions):
- January 2014: Google acquires the London-based AI lab DeepMind.
- June 2018: OpenAI launches GPT-1, the first iteration of the LLM that would lead to ChatGPT.
- February 2019: OpenAI launches GPT-2.
- July 2019: Microsoft invests $1bn in OpenAI.
- October 2019: Google launches Bidirectional Encoder Representations for Transformers for its search engine, making it easier for users to search using natural language.
- January 2020: Google unveils Meena, an AI-powered chatbot “that learns to respond sensibly to a given conversational context,” as described in a company blog post.
- May 2020: OpenAI launches GPT-3.
- June 2021: Google releases Language Model for Dialogue Applications (LaMDA), the second iteration of Meena.
- November 2021: Microsoft launches Azure OpenAI Service, making GPT-3 accessible to brands while simultaneously offering data security, compliance and other benefits.
- December 2021: DeepMind unveils Gopher, an LLM that “equals the performance of OpenAI’s GPT-3,” according to the company.
- January 2022: OpenAI releases GPT-3.5.
- May 2022: Google releases LaMDA 2. OpenAI introduces Gato, “a single generalist agent” that “can play Atari, caption images, chat, stack blocks with a real robot arm and much more,” according to a research paper published by the company.
- July 2022: Google fires Blake Lemoine, an engineer who claimed that LaMDA is conscious.
- November 2022: OpenAI releases ChatGPT.
- January 2023: Microsoft reportedly invests $10bn in OpenAI. ChatGPT reaches 100 million monthly active users, becoming the fastest-growing consumer product in history.
- February 2023: Microsoft introduces AI-powered upgrades to its Bing and Edge platforms.
- March 2023: OpenAI releases GPT-4. Google releases Bard, an AI-powered chatbot, which evolved out of LaMDA and was developed in response to the success of ChatGPT.
- April 2023: The Future of Life Institute publishes its open letter calling upon “all AI labs to immediately pause for at least six months the training of AI systems more powerful than GPT-4.”
Source: The Drum
Recent Comments