How, why and when generative AI will kill the search engine

The big news today is Google Bard vs. ChatGPT-infused Bing. But the real news is that the search engine is dead.

Self portraits created by ChatGPT describing itself physically, then entering that description as prompts into Midjourney via human Simon B.

Google and Microsoft just introduced plans to use generative AI in their respective search engines. But why bother? When AI is baked into everything, who needs search engines?

Google’s rumored ChatGPT competitor becomes available soon to “trusted testers,” then to the untrusted public “in the coming weeks.” (Whether they mean two weeks or two thousand weeks, we cannot know.)

Google’s “Bard” branding — not to be confused with the “Google Barge” — is another spectacular demonstration of Google’s incompetence with branding. (In fairness, ChatGPT is worse.)

“Bard” is presumably a reference to William Shakespeare, the kind of linguistic genius technologies like “Bard” will put out of business.

Bard is based on Lambda, the AI tech one Google engineer got fired over for falsely saying that it’s sentient.

I’ll let ChatGPT summarize the technology and purpose behind itself:

“ChatGPT is based on the Transformer architecture and is built using the GPT-3 (Generative Pretrained Transformer 3) model, developed by OpenAI. The model is trained on a massive amount of diverse text data, allowing it to generate human-like responses to a wide range of questions. It uses deep learning techniques, including neural networks and attention mechanisms, to process and generate text.”

Then I asked how ChatGPT and Google’s Lamda are different, and it wrote:

“Google Lambda is a machine learning model developed by Google, while ChatGPT is based on the GPT-3 model developed by OpenAI. Both models are designed to generate text, but their approach and functionality differ in some important ways. For example, Lambda is designed to generate a response to a prompt, while ChatGPT is designed to generate a response based on the input it receives. Additionally, Lambda has been trained on a specific domain (such as customer service), while ChatGPT has been trained on a diverse range of text data, making it more versatile. In summary, while both Google Lambda and ChatGPT use similar deep learning techniques to generate text, they have different training data, architectures, and use cases.”

Is this right? Who knows? ChatGPT is famous for spouting perfectly constructed but bland falsehoods with confidence.

“Bard seeks to combine the breadth of the world's knowledge with the power, intelligence, and creativity of our large language models," wrote Google chief Sundar Pichai. (If it’s so great, why didn’t Pichai dogfood it by letting Bard write the post?)

Pichai went on to say he wanted Google’s AI to be “bold and responsible” and presumably as capable as Pichai himself at blandly covering all the bases while neither exciting nor offending anyone. Bard will also be like Pichai in that it will use a “lightweight” version of Lamda (to save money on the electric bill, or something).

Of course, Google is freaking out because people are using ChatGPT as a search engine.

Microsoft made a preemptive strike against Google Bard today by announcing a new version of its Bing search engine and Edge browser accompanied (as rumored) a more advanced version of ChatGPT (based on GPT-4 rather than GPT-3) than we’ve all come to know and loath. Bing will sport a new “chat” feature for interacting with ChatGPT.

Microsoft is the main owner of ChatGPT. But legal and technical ownership by OpenAI (which in fact owns a 2% interest) gives Microsoft the legal and public relations cover to use AI in its products without itself being tarnished by ChatGPT transgressions.

I don’t think Google has the stomach for the kind of intellectual property theft, wrong answers and commentary on controversial subjects (like religion) that makes ChatGPT so fun. And they definitely don’t have any desire to wind up in court the way OpenAI increasingly is. Google won’t be happy when people use Bard to make and smuggle drugs or efficiently and automatically commit cyber crimes, for example.

I also think Google would freak out if users “jailbroke” Bard the way they’re doing with ChatGPT, tricking it into violating its own rules.

In any event, I think Google’s Bard will barely make a ripple. Dozens of competitors will be coming out of the woodwork. Users in the “free world” will have access to various ChatGPT-like AI all day, every day. But the bots will also divide us globally.

Baidu’s “Ernie Bot,” which will be a lot like ChatGPT, should go public in March. It will be the first authoritarian AI, with the ability to censor content and generate pro-Beijing propaganda on behalf of the Chinese Communist Party. Look for Orwellian Russian and North Korean bots in the near future, and later bots that give users a religious point of view coming out of Iran, Saudi Arabia and elsewhere. Ignorance is strength!

Authoritarian governments will ban foreign bots and require their own. Generative AI tools will eventually come to replace religion and political propaganda in school curricula as the main tools for the government indoctrination of children.

The trouble with building generative AI into search engines is that other AI tools will be baked into everything from email platforms to spreadsheets to cloud storage services to everyday smartphone apps. Major professional tools, including financial platforms, real estate tools, medical platforms and more will all come to integrate generative AI tools.

(There are already more than 1,500 tools integrating generative AI.)

The pièce de résistance, of course, will be AI in voice-based assistant platforms like Siri, Alexa and, of course Google Voice, which will be central to the future of augmented and virtual reality platforms.

When AI’s answers are everywhere, why visit a search engine?

The big story isn’t little announcements by Google and Microsoft.

The big story is that generative AI tools like ChatGPT will become ubiquitous and transform human culture faster than anything ever has, starting now. A few years from now, we’ll have to use AI if we want to remember what search engines were.

---

Mike’s List of Brilliantly Bad Ideas

1. Keyboard enthusiast builds the giant keyboard of his dreams

A keyboard-enthusiast maker and YouTuber who goes by the name “Glarses” fell in love with a giant, fully-functional mechanical keyboard demoed at CES 2018 by Razor. They wouldn’t sell it to him. So he built his own. Stood on its side, the keyboard is as tall as Glarses. And it cost $14,000 to build. Worth it!

2. Always be ready to brainstorm with this whiteboard wallet!

Silicon Valley runs on white boards. Trouble is, when you’re a digital nomad and always on the move, where do you put the white board? The Memo White Board Wallets says: in your pants! (The wallet holds your amazing startup ideas — and even your cash and cards!

3. A functional vending machine backpack? Why, Japan? Why?

Gacha Gacha Machines are toy-dispensing vending machines that have been popular in Japan since the 1960s. Now, Takara Tomy Arts is releasing a Gacha packpack. It even functions like a vending machine, and dispenses capsules (included with the backpack) filled with the toys of your choice.

---

Mike’s List of Shameless Self Promotions

1. 6 surprising facts about ChatGPT nobody told you

2. How the Silk Road Affair changed law enforcement

3. Hybrid-work hardware debuts at CES. Wait…, what?

4. Why Zero Trust works when everything else doesn’t

5. Synthetic media creates new social engineering threats

6. What to know about the Pentagon’s new push for zero trust

---

Where I’m at: Silicon Valley, California