• The Prohuman
  • Posts
  • AI Can Read Molecules But Can It Predict Results?

AI Can Read Molecules But Can It Predict Results?

Plus: Superhuman Makes Its AI Assistant More Personal

In partnership with

Hello, Prohuman

Today, we will talk about these stories:

  • OpenAI Tests AI’s Ability to Choose Drug Candidates

  • Superhuman Wants Go Across Your Work Apps

  • Google's AI Answers Are Changing Search Advertising

Frontier models perform better in Attio’s agent harness.

We gave GPT-6 Astra the same job twice: analyze 100 sales calls and draft the follow-ups, once in Ask Attio and once in ChatGPT.

Ask Attio finished in 4 minutes 49 seconds. ChatGPT took 26 minutes 11 seconds, about 5.4 times as long.

The most ambitious GTM work requires more than frontier intelligence. It requires architecture that can put that intelligence to work at scale.

AI Can Analyze Molecules, but Choosing Drugs Is Harder

Image credits: OpenAI

OpenAI's latest drug research test reveals a gap between understanding molecules and predicting what they'll actually do.

On November 20, OpenAI introduced MedChemBench, a benchmark built from 99 medicinal chemistry problems involving molecular structures, experimental results, and drug candidate selection.

GPT-6.2 Astra scored 84.5%, showing strong performance on tasks that normally require scientists to examine chemical drawings and interpret laboratory evidence.

But there's a catch.

The models performed much worse when asked to select promising compounds or predict their measured properties, earning roughly half the available credit in those areas.

That matters because drug researchers need to know how a compound behaves in the body, including whether it can enter cells or breaks down too quickly.

A correct chemical structure helps. A reliable prediction could influence which compound scientists spend money testing next.

My read is that AI is becoming useful for reviewing existing medicinal chemistry research, while its ability to guide new experiments still needs stronger evidence.

The real question is whether these scores will hold up when scientists test compounds the models have never encountered.

Superhuman Wants Go to Handle More of Your Work

image credits: Superhuman

Superhuman wants its Go assistant to understand your work well enough to act before you ask.

The company is expanding Go across Docs, Mail, and connected workplace apps, giving it access to emails, calendars, writing preferences, and company information.

Go can now draft replies, organize inboxes, launch background tasks, and use 20 new connected agents, including DeepL, Amplitude, and Databricks.

It gets personal.

The assistant works alongside your cursor, learns how you write, and can even turn meeting notes into a first proposal draft.

Superhuman says Go already has over 1 million active users, which gives the company a meaningful starting point as more AI assistants compete for daily workplace use.

I think the strongest feature here is shared context across apps, since employees often spend time moving information between email, documents, and meetings.

But connecting more workplace data also raises questions about permissions and how much autonomy people will comfortably give these tools.

Will users trust Go to make decisions across their work apps without checking every action first?

Google's AI Search Is Putting Pressure on Ad Revenue

image credits: Google

Google's AI-generated answers are pushing traditional search ads further down the page, and some advertisers are starting to spend less.

One marketer cut a client's monthly Google search budget from $2,000 to $500, redirecting money toward platforms like Tubi, Bing, and Reddit.

Google's search advertising revenue still grew 12% last quarter, although that was below Wall Street's expectations.

The problem is visibility.

When people get answers directly from AI Overviews or continue chatting in AI Mode, they may have fewer reasons to click an ad or visit a business website.

I think the bigger concern for Google is that advertisers are losing some control over where their money goes and whether customers actually see their promotions.

Google says AI is increasing search activity, but more searches won't automatically mean better returns for advertisers, especially when AI ad placements remain unpredictable.

Some businesses are already testing alternatives, and that could gradually shift advertising budgets toward competing platforms.

How much longer will advertisers keep paying Google if its AI answers reduce the clicks they're buying?

Prohuman team

Covers emerging technology, AI models, and the people building the next layer of the internet.

Founder

Writes about how new interfaces, reasoning models, and automation are reshaping human work.

Founder

Free Guides

Explore our free guides and products to get into AI and master it.

All of them are free to access and would stay free for you.

Feeling generous?

You know someone who loves breakthroughs as much as you do.

Share The Prohuman it’s how smart people stay one update ahead.