Google’s latest announcement could be a
threat to OpenAI’s future. Google wants to lead the AI race and overtake
ChatGPT. Their latest offering, combining search and
AI, is creating a massive buzz. Will they be able to leave their Microsoft-backed
rivals in the dust? The potential for search engine AI integration
is massive. Google’s new tool could make search more
conversational, contextual, and creative.
Let’s find out how worried OpenAI should be about their closest rivals. There are plenty of players in the AI industry. But despite this, it sometimes seems like
a two horse race. The battle of the Iarge language models has
been between OpenAI and Google. So far, OpenAI’s ChatGPT has been in the
lead. Especially after the transition to GPT-4. Google’s Bard, so far, has not been able
to compete. But Google is a massive company with plenty
of resources. And they don’t want to lag behind in the
AI battle forever. They are clearly pulling out all the stops
to take the lead. And their latest product may finally allow
them to do just that.
Now, Google became a tech titan because of its search engine. So it makes sense that they are returning
to this area. It’s what they do best, and it’s a natural
application for AI. There is a lot of potential for AI integration
in search engines. We’ve already seen this with Microsoft’s
Bing, which is powered by ChatGPT. Google has dominated over Bing as a search
engine. But with the arrival of ChatGPT, this dominance
has been threatened. ChatGPT gave Bing a boost, allowing it to
exceed 100 million users. Samsung even considered replacing Google with
Bing as their phone’s default browser. Ultimately, they stuck with Google. But they wouldn’t even be considering this
change without ChatGPT’s success. It seems that Google has noticed that Bing
is growing in popularity. Because they’ve just announced the launch
of a new search tool. And it has the potential to be a major gamechanger. Google recently announced the beginning of
a new era of AI-powered search. They’re calling it Search Generative Experience
(SGE). This involves incorporating generative AI
to enhance the search experience. This could completely change how we interact
with a search engine. Right now, it’s simply feeding data and
getting results. But with AI, it could be far more conversational,
contextual, and creative. Firstly, SGE allows users to interact with
Google using natural language. Secondly, you can receive answers to queries
without the need for specific keywords. With SGE, Google can understand the context
of your query. It can provide personalized results based
on location, preferences, and history. But the most exciting feature of SGE is its
capacity for generative AI. This is what we’ve seen with ChatGPT and
other large language models. It’s AI that has the ability, for the first
time, to create. With SGE, you’re able to use a search
engine to create content. This could be images, poems, blog posts, the
potential is pretty limitless. It’s important to note that AI used in search
is nothing new. Google has been integrating AI into its search
engine for years. It started with machine learning applications
such as spelling correction. Then there have been more recent breakthroughs
in AI. These include Bidirectional Encoder Representations
from Transformers (BERT). Also, the Multitask Unified Model (MUM). Both of these have improved search quality
and understanding of user queries. Another important issue to address is what
AI technology Google is using. You may wonder if SGE is powered by Bard. This is Google’s most famous large language
model. But, it has lagged behind ChatGPT in many
areas. It may, then, relieve some people that SGE
will not use Bard. Rather, it will use PaLM2, a game-changing
LLM we’ve spoken about on this channel. It will be trained on a dataset called Infiniset. There’s also the issue of sources and attribution.
Bard has been controversial for not citing sources in its answers. This has proven to be an unpopular move. Many content creators fear losing visibility
and traffic for their work. Google has come up with a response to this. SGE searches will not rely solely on AI-generated
answers. Instead, three featured snippets will feature
alongside the AI-generated response. These snippets are chosen based on their closeness
to the AI-generated answer. Additionally, users can access further sources. There’s a button in the upper right corner,
which provides additional citations. Google has not yet announced an official launch
date for SGE. For now, there are a few ways it can be accessed. You can use Search Labs. This is an invite-only program allowing users
to try out new features. Or, you can use Performance Max. This is a campaign type advertisers will find
in Google Ads. Both of these offer you a way of experimenting
with this technology. Which, we hope, will be available to the broader
public fairly soon. SGE will also be accessible on Chrome desktop
in the US. Google says that for now they want to gather
user feedback. The aim being to improve the SE experience
over time. Google is not without competition when it
comes to AI and search. Of course, there’s Microsoft Bing, which
already uses ChatGPT. But there’s also a lesser-known competitor. Vectara, a startup, is an emerging player
in the AI search space. They claim to offer a better generative AI
conversational search platform. Then there’s Elasticsearch, an open-source
search engine. Despite all the competition, though, Google’s
in a great position to dominate. They have brand recognition, funding, and
know-how. Everything you’d need to own the AI search
space. The announcement of SGE has generated significant
hype. Many expect it to revolutionize the way we
use search. Will it deliver on its promise? Only time will tell. But if it does, it could be very bad news
for OpenAI. They may end up as a mere footprint in the
history books. What improvements will SGE bring to search
engines? We don’t know, but there are plenty of clues. This is what Google said about it in a blog
post: SGE aims to improve search engine efficiency. Through AI, the aim is to make searching less
work. Users should be able to understand a topic
faster. They will be able to uncover new viewpoints
and insights. And, they will be able to get things done
more easily. The new search engine will also feature AI-powered
snapshots. These will be snapshots of key information
related to a search query. They will include links for users to dig deeper
and explore further. SGE will also come with a conversational mode. Users will have the ability to ask follow-up
questions. Context from previous questions will be carried
over to the next search. This should help users continue their exploration
naturally. Another major feature of SGE is its ability
to provide shopping assistance. The generative AI capabilities will assist
users in making shopping decisions. It will be able to provide a comprehensive
view of products. This includes noteworthy factors to consider,
product descriptions with reviews, ratings, prices, and images. The shopping experience will be built on Google's
Shopping Graph. This contains a vast dataset of constantly-changing
products and information. Google hopes SGE will introduce a wider range
of voices and sources. Google aims to highlight and drive attention
to content on the web. This will provide users with insights from
various sources. This ensures users have access to diverse
perspectives. It will mean a less biased search. And, you should have access to more information
to inform your decisions. The tech giant also says they’re taking
a responsible approach to SGE. In the blog, Google acknowledges the limitations
of generative AI. Search engines rely on fine-tuned systems. For this reason, a responsible approach is
vital. Google says they’re focusing on quality
and taking a deliberate, responsible approach. They’re also putting in some breaks and
safety features. For example, only certain types of queries
will lead to AI generated results. What kind of queries won’t involve AI? We took a look at reviews from those who have
tried SGE. There are some queries that don’t include
SGE responses. These include sensitive topics, and ‘your
money or your life’ (YMYL) searches. YMYL is when you search for something that
Google deems important. Or, in Google’s own words: Anything which
impacts “the future happiness, health, financial stability, or safety of users”. SGE results feature a disclaimer acknowledging
the experimental nature of generative AI. Feedback so far has been mainly positive. However, reviewers found that SGE was lacking
in some areas. SGE did not respond or failed to generate
a response in some cases. In other cases, the generated responses contained
errors. These included outdated information or incomplete
answers. SGE's accuracy and completeness varied across
different queries. Some responses are highly accurate and complete. Others contain errors and omissions. For Google’s part, they’re taking pains
not to oversell SGE. They’ve emphasized caution with generative
AI and acknowledged that mistakes can occur. But despite limitations and occasional errors,
SGE shows plenty of promise. Google will keep refining and improving on
it. So it’s set to get a whole lot better. So it’s possible that in a few years we’ll
all be using it. And, we may have nothing but memories when
someone brings up OpenAI.
